diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6026 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 7508,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001,
+      "learning_rate": 8.50153746173217e-06,
+      "loss": 0.7923,
+      "r_loss": 0.0,
+      "step": 10,
+      "steps": 10
+    },
+    {
+      "epoch": 0.003,
+      "learning_rate": 1.1057744947017939e-05,
+      "loss": 0.7252,
+      "r_loss": 0.0,
+      "step": 20,
+      "steps": 20
+    },
+    {
+      "epoch": 0.004,
+      "learning_rate": 1.255303046997284e-05,
+      "loss": 0.7285,
+      "r_loss": 0.0,
+      "step": 30,
+      "steps": 30
+    },
+    {
+      "epoch": 0.005,
+      "learning_rate": 1.3613952432303706e-05,
+      "loss": 0.7204,
+      "r_loss": 0.0,
+      "step": 40,
+      "steps": 40
+    },
+    {
+      "epoch": 0.007,
+      "learning_rate": 1.4436867438178569e-05,
+      "loss": 0.7221,
+      "r_loss": 0.0,
+      "step": 50,
+      "steps": 50
+    },
+    {
+      "epoch": 0.008,
+      "learning_rate": 1.5109237955258607e-05,
+      "loss": 0.7153,
+      "r_loss": 0.0,
+      "step": 60,
+      "steps": 60
+    },
+    {
+      "epoch": 0.009,
+      "learning_rate": 1.5677719127349665e-05,
+      "loss": 0.6948,
+      "r_loss": 0.0,
+      "step": 70,
+      "steps": 70
+    },
+    {
+      "epoch": 0.011,
+      "learning_rate": 1.6170159917589475e-05,
+      "loss": 0.7023,
+      "r_loss": 0.0,
+      "step": 80,
+      "steps": 80
+    },
+    {
+      "epoch": 0.012,
+      "learning_rate": 1.660452347821351e-05,
+      "loss": 0.7053,
+      "r_loss": 0.0,
+      "step": 90,
+      "steps": 90
+    },
+    {
+      "epoch": 0.013,
+      "learning_rate": 1.6993074923464342e-05,
+      "loss": 0.6908,
+      "r_loss": 0.0,
+      "step": 100,
+      "steps": 100
+    },
+    {
+      "epoch": 0.015,
+      "learning_rate": 1.7344562460127094e-05,
+      "loss": 0.7005,
+      "r_loss": 0.0,
+      "step": 110,
+      "steps": 110
+    },
+    {
+      "epoch": 0.016,
+      "learning_rate": 1.7665445440544375e-05,
+      "loss": 0.6864,
+      "r_loss": 0.0,
+      "step": 120,
+      "steps": 120
+    },
+    {
+      "epoch": 0.017,
+      "learning_rate": 1.796062916809319e-05,
+      "loss": 0.7273,
+      "r_loss": 0.0,
+      "step": 130,
+      "steps": 130
+    },
+    {
+      "epoch": 0.019,
+      "learning_rate": 1.823392661263543e-05,
+      "loss": 0.7019,
+      "r_loss": 0.0,
+      "step": 140,
+      "steps": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.848836044641924e-05,
+      "loss": 0.6934,
+      "r_loss": 0.0,
+      "step": 150,
+      "steps": 150
+    },
+    {
+      "epoch": 0.021,
+      "learning_rate": 1.8726367402875247e-05,
+      "loss": 0.6908,
+      "r_loss": 0.0,
+      "step": 160,
+      "steps": 160
+    },
+    {
+      "epoch": 0.023,
+      "learning_rate": 1.8949940572363726e-05,
+      "loss": 0.677,
+      "r_loss": 0.0,
+      "step": 170,
+      "steps": 170
+    },
+    {
+      "epoch": 0.024,
+      "learning_rate": 1.916073096349928e-05,
+      "loss": 0.6889,
+      "r_loss": 0.0,
+      "step": 180,
+      "steps": 180
+    },
+    {
+      "epoch": 0.025,
+      "learning_rate": 1.936012156854803e-05,
+      "loss": 0.6831,
+      "r_loss": 0.0,
+      "step": 190,
+      "steps": 190
+    },
+    {
+      "epoch": 0.027,
+      "learning_rate": 1.9549282408750108e-05,
+      "loss": 0.6893,
+      "r_loss": 0.0,
+      "step": 200,
+      "steps": 200
+    },
+    {
+      "epoch": 0.028,
+      "learning_rate": 1.9729212135590334e-05,
+      "loss": 0.684,
+      "r_loss": 0.0,
+      "step": 210,
+      "steps": 210
+    },
+    {
+      "epoch": 0.029,
+      "learning_rate": 1.9900769945412862e-05,
+      "loss": 0.7032,
+      "r_loss": 0.0,
+      "step": 220,
+      "steps": 220
+    },
+    {
+      "epoch": 0.031,
+      "learning_rate": 1.9991764625102995e-05,
+      "loss": 0.6922,
+      "r_loss": 0.0,
+      "step": 230,
+      "steps": 230
+    },
+    {
+      "epoch": 0.032,
+      "learning_rate": 1.9964313375446307e-05,
+      "loss": 0.6779,
+      "r_loss": 0.0,
+      "step": 240,
+      "steps": 240
+    },
+    {
+      "epoch": 0.033,
+      "learning_rate": 1.993686212578962e-05,
+      "loss": 0.6851,
+      "r_loss": 0.0,
+      "step": 250,
+      "steps": 250
+    },
+    {
+      "epoch": 0.035,
+      "learning_rate": 1.990941087613293e-05,
+      "loss": 0.6834,
+      "r_loss": 0.0,
+      "step": 260,
+      "steps": 260
+    },
+    {
+      "epoch": 0.036,
+      "learning_rate": 1.9881959626476245e-05,
+      "loss": 0.68,
+      "r_loss": 0.0,
+      "step": 270,
+      "steps": 270
+    },
+    {
+      "epoch": 0.037,
+      "learning_rate": 1.9854508376819557e-05,
+      "loss": 0.674,
+      "r_loss": 0.0,
+      "step": 280,
+      "steps": 280
+    },
+    {
+      "epoch": 0.039,
+      "learning_rate": 1.982705712716287e-05,
+      "loss": 0.6785,
+      "r_loss": 0.0,
+      "step": 290,
+      "steps": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9799605877506184e-05,
+      "loss": 0.6976,
+      "r_loss": 0.0,
+      "step": 300,
+      "steps": 300
+    },
+    {
+      "epoch": 0.041,
+      "learning_rate": 1.9772154627849495e-05,
+      "loss": 0.6639,
+      "r_loss": 0.0,
+      "step": 310,
+      "steps": 310
+    },
+    {
+      "epoch": 0.043,
+      "learning_rate": 1.9744703378192804e-05,
+      "loss": 0.6823,
+      "r_loss": 0.0,
+      "step": 320,
+      "steps": 320
+    },
+    {
+      "epoch": 0.044,
+      "learning_rate": 1.971725212853612e-05,
+      "loss": 0.6834,
+      "r_loss": 0.0,
+      "step": 330,
+      "steps": 330
+    },
+    {
+      "epoch": 0.045,
+      "learning_rate": 1.968980087887943e-05,
+      "loss": 0.6843,
+      "r_loss": 0.0,
+      "step": 340,
+      "steps": 340
+    },
+    {
+      "epoch": 0.047,
+      "learning_rate": 1.9662349629222742e-05,
+      "loss": 0.678,
+      "r_loss": 0.0,
+      "step": 350,
+      "steps": 350
+    },
+    {
+      "epoch": 0.048,
+      "learning_rate": 1.9634898379566057e-05,
+      "loss": 0.6742,
+      "r_loss": 0.0,
+      "step": 360,
+      "steps": 360
+    },
+    {
+      "epoch": 0.049,
+      "learning_rate": 1.960744712990937e-05,
+      "loss": 0.6652,
+      "r_loss": 0.0,
+      "step": 370,
+      "steps": 370
+    },
+    {
+      "epoch": 0.051,
+      "learning_rate": 1.957999588025268e-05,
+      "loss": 0.6714,
+      "r_loss": 0.0,
+      "step": 380,
+      "steps": 380
+    },
+    {
+      "epoch": 0.052,
+      "learning_rate": 1.9552544630595992e-05,
+      "loss": 0.6664,
+      "r_loss": 0.0,
+      "step": 390,
+      "steps": 390
+    },
+    {
+      "epoch": 0.053,
+      "learning_rate": 1.9525093380939303e-05,
+      "loss": 0.6661,
+      "r_loss": 0.0,
+      "step": 400,
+      "steps": 400
+    },
+    {
+      "epoch": 0.055,
+      "learning_rate": 1.9497642131282615e-05,
+      "loss": 0.6736,
+      "r_loss": 0.0,
+      "step": 410,
+      "steps": 410
+    },
+    {
+      "epoch": 0.056,
+      "learning_rate": 1.947019088162593e-05,
+      "loss": 0.6731,
+      "r_loss": 0.0,
+      "step": 420,
+      "steps": 420
+    },
+    {
+      "epoch": 0.057,
+      "learning_rate": 1.944273963196924e-05,
+      "loss": 0.6613,
+      "r_loss": 0.0,
+      "step": 430,
+      "steps": 430
+    },
+    {
+      "epoch": 0.059,
+      "learning_rate": 1.9415288382312553e-05,
+      "loss": 0.6651,
+      "r_loss": 0.0,
+      "step": 440,
+      "steps": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9387837132655865e-05,
+      "loss": 0.6725,
+      "r_loss": 0.0,
+      "step": 450,
+      "steps": 450
+    },
+    {
+      "epoch": 0.061,
+      "learning_rate": 1.9363131007964846e-05,
+      "loss": 0.6623,
+      "r_loss": 0.0,
+      "step": 460,
+      "steps": 460
+    },
+    {
+      "epoch": 0.063,
+      "learning_rate": 1.933567975830816e-05,
+      "loss": 0.6758,
+      "r_loss": 0.0,
+      "step": 470,
+      "steps": 470
+    },
+    {
+      "epoch": 0.064,
+      "learning_rate": 1.930822850865147e-05,
+      "loss": 0.6611,
+      "r_loss": 0.0,
+      "step": 480,
+      "steps": 480
+    },
+    {
+      "epoch": 0.065,
+      "learning_rate": 1.928077725899478e-05,
+      "loss": 0.6701,
+      "r_loss": 0.0,
+      "step": 490,
+      "steps": 490
+    },
+    {
+      "epoch": 0.067,
+      "learning_rate": 1.9253326009338096e-05,
+      "loss": 0.6657,
+      "r_loss": 0.0,
+      "step": 500,
+      "steps": 500
+    },
+    {
+      "epoch": 0.068,
+      "learning_rate": 1.9225874759681408e-05,
+      "loss": 0.6668,
+      "r_loss": 0.0,
+      "step": 510,
+      "steps": 510
+    },
+    {
+      "epoch": 0.069,
+      "learning_rate": 1.919842351002472e-05,
+      "loss": 0.6737,
+      "r_loss": 0.0,
+      "step": 520,
+      "steps": 520
+    },
+    {
+      "epoch": 0.071,
+      "learning_rate": 1.9170972260368035e-05,
+      "loss": 0.6586,
+      "r_loss": 0.0,
+      "step": 530,
+      "steps": 530
+    },
+    {
+      "epoch": 0.072,
+      "learning_rate": 1.9143521010711346e-05,
+      "loss": 0.6704,
+      "r_loss": 0.0,
+      "step": 540,
+      "steps": 540
+    },
+    {
+      "epoch": 0.073,
+      "learning_rate": 1.9116069761054658e-05,
+      "loss": 0.6573,
+      "r_loss": 0.0,
+      "step": 550,
+      "steps": 550
+    },
+    {
+      "epoch": 0.075,
+      "learning_rate": 1.908861851139797e-05,
+      "loss": 0.6558,
+      "r_loss": 0.0,
+      "step": 560,
+      "steps": 560
+    },
+    {
+      "epoch": 0.076,
+      "learning_rate": 1.906116726174128e-05,
+      "loss": 0.6563,
+      "r_loss": 0.0,
+      "step": 570,
+      "steps": 570
+    },
+    {
+      "epoch": 0.077,
+      "learning_rate": 1.9033716012084593e-05,
+      "loss": 0.6609,
+      "r_loss": 0.0,
+      "step": 580,
+      "steps": 580
+    },
+    {
+      "epoch": 0.079,
+      "learning_rate": 1.9006264762427908e-05,
+      "loss": 0.6551,
+      "r_loss": 0.0,
+      "step": 590,
+      "steps": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.897881351277122e-05,
+      "loss": 0.6599,
+      "r_loss": 0.0,
+      "step": 600,
+      "steps": 600
+    },
+    {
+      "epoch": 0.081,
+      "learning_rate": 1.895136226311453e-05,
+      "loss": 0.6557,
+      "r_loss": 0.0,
+      "step": 610,
+      "steps": 610
+    },
+    {
+      "epoch": 0.083,
+      "learning_rate": 1.8926656138423513e-05,
+      "loss": 0.6588,
+      "r_loss": 0.0,
+      "step": 620,
+      "steps": 620
+    },
+    {
+      "epoch": 0.084,
+      "learning_rate": 1.8899204888766824e-05,
+      "loss": 0.6637,
+      "r_loss": 0.0,
+      "step": 630,
+      "steps": 630
+    },
+    {
+      "epoch": 0.085,
+      "learning_rate": 1.8871753639110136e-05,
+      "loss": 0.6464,
+      "r_loss": 0.0,
+      "step": 640,
+      "steps": 640
+    },
+    {
+      "epoch": 0.087,
+      "learning_rate": 1.8844302389453447e-05,
+      "loss": 0.6582,
+      "r_loss": 0.0,
+      "step": 650,
+      "steps": 650
+    },
+    {
+      "epoch": 0.088,
+      "learning_rate": 1.881685113979676e-05,
+      "loss": 0.6457,
+      "r_loss": 0.0,
+      "step": 660,
+      "steps": 660
+    },
+    {
+      "epoch": 0.089,
+      "learning_rate": 1.8789399890140074e-05,
+      "loss": 0.6598,
+      "r_loss": 0.0,
+      "step": 670,
+      "steps": 670
+    },
+    {
+      "epoch": 0.091,
+      "learning_rate": 1.8761948640483386e-05,
+      "loss": 0.65,
+      "r_loss": 0.0,
+      "step": 680,
+      "steps": 680
+    },
+    {
+      "epoch": 0.092,
+      "learning_rate": 1.8734497390826697e-05,
+      "loss": 0.6348,
+      "r_loss": 0.0,
+      "step": 690,
+      "steps": 690
+    },
+    {
+      "epoch": 0.093,
+      "learning_rate": 1.8707046141170012e-05,
+      "loss": 0.6596,
+      "r_loss": 0.0,
+      "step": 700,
+      "steps": 700
+    },
+    {
+      "epoch": 0.095,
+      "learning_rate": 1.8679594891513324e-05,
+      "loss": 0.641,
+      "r_loss": 0.0,
+      "step": 710,
+      "steps": 710
+    },
+    {
+      "epoch": 0.096,
+      "learning_rate": 1.8652143641856632e-05,
+      "loss": 0.6598,
+      "r_loss": 0.0,
+      "step": 720,
+      "steps": 720
+    },
+    {
+      "epoch": 0.097,
+      "learning_rate": 1.8624692392199947e-05,
+      "loss": 0.6606,
+      "r_loss": 0.0,
+      "step": 730,
+      "steps": 730
+    },
+    {
+      "epoch": 0.099,
+      "learning_rate": 1.859724114254326e-05,
+      "loss": 0.6471,
+      "r_loss": 0.0,
+      "step": 740,
+      "steps": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.856978989288657e-05,
+      "loss": 0.6613,
+      "r_loss": 0.0,
+      "step": 750,
+      "steps": 750
+    },
+    {
+      "epoch": 0.101,
+      "learning_rate": 1.8542338643229886e-05,
+      "loss": 0.6495,
+      "r_loss": 0.0,
+      "step": 760,
+      "steps": 760
+    },
+    {
+      "epoch": 0.103,
+      "learning_rate": 1.8514887393573197e-05,
+      "loss": 0.6577,
+      "r_loss": 0.0,
+      "step": 770,
+      "steps": 770
+    },
+    {
+      "epoch": 0.104,
+      "learning_rate": 1.848743614391651e-05,
+      "loss": 0.6466,
+      "r_loss": 0.0,
+      "step": 780,
+      "steps": 780
+    },
+    {
+      "epoch": 0.105,
+      "learning_rate": 1.845998489425982e-05,
+      "loss": 0.655,
+      "r_loss": 0.0,
+      "step": 790,
+      "steps": 790
+    },
+    {
+      "epoch": 0.107,
+      "learning_rate": 1.8432533644603132e-05,
+      "loss": 0.6291,
+      "r_loss": 0.0,
+      "step": 800,
+      "steps": 800
+    },
+    {
+      "epoch": 0.108,
+      "learning_rate": 1.8405082394946444e-05,
+      "loss": 0.6574,
+      "r_loss": 0.0,
+      "step": 810,
+      "steps": 810
+    },
+    {
+      "epoch": 0.109,
+      "learning_rate": 1.837763114528976e-05,
+      "loss": 0.6576,
+      "r_loss": 0.0,
+      "step": 820,
+      "steps": 820
+    },
+    {
+      "epoch": 0.111,
+      "learning_rate": 1.835017989563307e-05,
+      "loss": 0.646,
+      "r_loss": 0.0,
+      "step": 830,
+      "steps": 830
+    },
+    {
+      "epoch": 0.112,
+      "learning_rate": 1.8322728645976382e-05,
+      "loss": 0.6479,
+      "r_loss": 0.0,
+      "step": 840,
+      "steps": 840
+    },
+    {
+      "epoch": 0.113,
+      "learning_rate": 1.8295277396319694e-05,
+      "loss": 0.6435,
+      "r_loss": 0.0,
+      "step": 850,
+      "steps": 850
+    },
+    {
+      "epoch": 0.115,
+      "learning_rate": 1.8267826146663005e-05,
+      "loss": 0.6473,
+      "r_loss": 0.0,
+      "step": 860,
+      "steps": 860
+    },
+    {
+      "epoch": 0.116,
+      "learning_rate": 1.824037489700632e-05,
+      "loss": 0.6554,
+      "r_loss": 0.0,
+      "step": 870,
+      "steps": 870
+    },
+    {
+      "epoch": 0.117,
+      "learning_rate": 1.8212923647349632e-05,
+      "loss": 0.6362,
+      "r_loss": 0.0,
+      "step": 880,
+      "steps": 880
+    },
+    {
+      "epoch": 0.119,
+      "learning_rate": 1.8185472397692944e-05,
+      "loss": 0.6489,
+      "r_loss": 0.0,
+      "step": 890,
+      "steps": 890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8158021148036255e-05,
+      "loss": 0.6496,
+      "r_loss": 0.0,
+      "step": 900,
+      "steps": 900
+    },
+    {
+      "epoch": 0.121,
+      "learning_rate": 1.8130569898379567e-05,
+      "loss": 0.6358,
+      "r_loss": 0.0,
+      "step": 910,
+      "steps": 910
+    },
+    {
+      "epoch": 0.123,
+      "learning_rate": 1.810311864872288e-05,
+      "loss": 0.6461,
+      "r_loss": 0.0,
+      "step": 920,
+      "steps": 920
+    },
+    {
+      "epoch": 0.124,
+      "learning_rate": 1.8075667399066194e-05,
+      "loss": 0.6464,
+      "r_loss": 0.0,
+      "step": 930,
+      "steps": 930
+    },
+    {
+      "epoch": 0.125,
+      "learning_rate": 1.8048216149409505e-05,
+      "loss": 0.6402,
+      "r_loss": 0.0,
+      "step": 940,
+      "steps": 940
+    },
+    {
+      "epoch": 0.127,
+      "learning_rate": 1.8020764899752817e-05,
+      "loss": 0.6423,
+      "r_loss": 0.0,
+      "step": 950,
+      "steps": 950
+    },
+    {
+      "epoch": 0.128,
+      "learning_rate": 1.799331365009613e-05,
+      "loss": 0.6318,
+      "r_loss": 0.0,
+      "step": 960,
+      "steps": 960
+    },
+    {
+      "epoch": 0.129,
+      "learning_rate": 1.796586240043944e-05,
+      "loss": 0.6353,
+      "r_loss": 0.0,
+      "step": 970,
+      "steps": 970
+    },
+    {
+      "epoch": 0.131,
+      "learning_rate": 1.793841115078275e-05,
+      "loss": 0.6379,
+      "r_loss": 0.0,
+      "step": 980,
+      "steps": 980
+    },
+    {
+      "epoch": 0.132,
+      "learning_rate": 1.7910959901126067e-05,
+      "loss": 0.6367,
+      "r_loss": 0.0,
+      "step": 990,
+      "steps": 990
+    },
+    {
+      "epoch": 0.133,
+      "learning_rate": 1.788350865146938e-05,
+      "loss": 0.6327,
+      "r_loss": 0.0,
+      "step": 1000,
+      "steps": 1000
+    },
+    {
+      "epoch": 0.135,
+      "learning_rate": 1.785605740181269e-05,
+      "loss": 0.6357,
+      "r_loss": 0.0,
+      "step": 1010,
+      "steps": 1010
+    },
+    {
+      "epoch": 0.136,
+      "learning_rate": 1.7828606152156e-05,
+      "loss": 0.6308,
+      "r_loss": 0.0,
+      "step": 1020,
+      "steps": 1020
+    },
+    {
+      "epoch": 0.137,
+      "learning_rate": 1.7801154902499313e-05,
+      "loss": 0.632,
+      "r_loss": 0.0,
+      "step": 1030,
+      "steps": 1030
+    },
+    {
+      "epoch": 0.139,
+      "learning_rate": 1.7773703652842628e-05,
+      "loss": 0.6369,
+      "r_loss": 0.0,
+      "step": 1040,
+      "steps": 1040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.774625240318594e-05,
+      "loss": 0.6308,
+      "r_loss": 0.0,
+      "step": 1050,
+      "steps": 1050
+    },
+    {
+      "epoch": 0.141,
+      "learning_rate": 1.771880115352925e-05,
+      "loss": 0.6442,
+      "r_loss": 0.0,
+      "step": 1060,
+      "steps": 1060
+    },
+    {
+      "epoch": 0.143,
+      "learning_rate": 1.7691349903872567e-05,
+      "loss": 0.6434,
+      "r_loss": 0.0,
+      "step": 1070,
+      "steps": 1070
+    },
+    {
+      "epoch": 0.144,
+      "learning_rate": 1.7663898654215878e-05,
+      "loss": 0.6484,
+      "r_loss": 0.0,
+      "step": 1080,
+      "steps": 1080
+    },
+    {
+      "epoch": 0.145,
+      "learning_rate": 1.7636447404559186e-05,
+      "loss": 0.638,
+      "r_loss": 0.0,
+      "step": 1090,
+      "steps": 1090
+    },
+    {
+      "epoch": 0.147,
+      "learning_rate": 1.76089961549025e-05,
+      "loss": 0.6337,
+      "r_loss": 0.0,
+      "step": 1100,
+      "steps": 1100
+    },
+    {
+      "epoch": 0.148,
+      "learning_rate": 1.7581544905245813e-05,
+      "loss": 0.652,
+      "r_loss": 0.0,
+      "step": 1110,
+      "steps": 1110
+    },
+    {
+      "epoch": 0.149,
+      "learning_rate": 1.7554093655589125e-05,
+      "loss": 0.6325,
+      "r_loss": 0.0,
+      "step": 1120,
+      "steps": 1120
+    },
+    {
+      "epoch": 0.151,
+      "learning_rate": 1.752664240593244e-05,
+      "loss": 0.6394,
+      "r_loss": 0.0,
+      "step": 1130,
+      "steps": 1130
+    },
+    {
+      "epoch": 0.152,
+      "learning_rate": 1.749919115627575e-05,
+      "loss": 0.6251,
+      "r_loss": 0.0,
+      "step": 1140,
+      "steps": 1140
+    },
+    {
+      "epoch": 0.153,
+      "learning_rate": 1.7471739906619063e-05,
+      "loss": 0.6304,
+      "r_loss": 0.0,
+      "step": 1150,
+      "steps": 1150
+    },
+    {
+      "epoch": 0.155,
+      "learning_rate": 1.7444288656962375e-05,
+      "loss": 0.6251,
+      "r_loss": 0.0,
+      "step": 1160,
+      "steps": 1160
+    },
+    {
+      "epoch": 0.156,
+      "learning_rate": 1.7416837407305686e-05,
+      "loss": 0.6169,
+      "r_loss": 0.0,
+      "step": 1170,
+      "steps": 1170
+    },
+    {
+      "epoch": 0.157,
+      "learning_rate": 1.7389386157648998e-05,
+      "loss": 0.6268,
+      "r_loss": 0.0,
+      "step": 1180,
+      "steps": 1180
+    },
+    {
+      "epoch": 0.158,
+      "learning_rate": 1.7361934907992313e-05,
+      "loss": 0.6305,
+      "r_loss": 0.0,
+      "step": 1190,
+      "steps": 1190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7334483658335625e-05,
+      "loss": 0.635,
+      "r_loss": 0.0,
+      "step": 1200,
+      "steps": 1200
+    },
+    {
+      "epoch": 0.161,
+      "learning_rate": 1.7307032408678936e-05,
+      "loss": 0.6142,
+      "r_loss": 0.0,
+      "step": 1210,
+      "steps": 1210
+    },
+    {
+      "epoch": 0.162,
+      "learning_rate": 1.7279581159022248e-05,
+      "loss": 0.6207,
+      "r_loss": 0.0,
+      "step": 1220,
+      "steps": 1220
+    },
+    {
+      "epoch": 0.164,
+      "learning_rate": 1.725212990936556e-05,
+      "loss": 0.6317,
+      "r_loss": 0.0,
+      "step": 1230,
+      "steps": 1230
+    },
+    {
+      "epoch": 0.165,
+      "learning_rate": 1.7224678659708874e-05,
+      "loss": 0.6434,
+      "r_loss": 0.0,
+      "step": 1240,
+      "steps": 1240
+    },
+    {
+      "epoch": 0.166,
+      "learning_rate": 1.7197227410052186e-05,
+      "loss": 0.6335,
+      "r_loss": 0.0,
+      "step": 1250,
+      "steps": 1250
+    },
+    {
+      "epoch": 0.168,
+      "learning_rate": 1.7169776160395498e-05,
+      "loss": 0.6341,
+      "r_loss": 0.0,
+      "step": 1260,
+      "steps": 1260
+    },
+    {
+      "epoch": 0.169,
+      "learning_rate": 1.714232491073881e-05,
+      "loss": 0.6226,
+      "r_loss": 0.0,
+      "step": 1270,
+      "steps": 1270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.711487366108212e-05,
+      "loss": 0.6135,
+      "r_loss": 0.0,
+      "step": 1280,
+      "steps": 1280
+    },
+    {
+      "epoch": 0.172,
+      "learning_rate": 1.7087422411425433e-05,
+      "loss": 0.6391,
+      "r_loss": 0.0,
+      "step": 1290,
+      "steps": 1290
+    },
+    {
+      "epoch": 0.173,
+      "learning_rate": 1.7059971161768748e-05,
+      "loss": 0.6315,
+      "r_loss": 0.0,
+      "step": 1300,
+      "steps": 1300
+    },
+    {
+      "epoch": 0.174,
+      "learning_rate": 1.703251991211206e-05,
+      "loss": 0.6224,
+      "r_loss": 0.0,
+      "step": 1310,
+      "steps": 1310
+    },
+    {
+      "epoch": 0.176,
+      "learning_rate": 1.700506866245537e-05,
+      "loss": 0.6376,
+      "r_loss": 0.0,
+      "step": 1320,
+      "steps": 1320
+    },
+    {
+      "epoch": 0.177,
+      "learning_rate": 1.6977617412798683e-05,
+      "loss": 0.6263,
+      "r_loss": 0.0,
+      "step": 1330,
+      "steps": 1330
+    },
+    {
+      "epoch": 0.178,
+      "learning_rate": 1.6950166163141994e-05,
+      "loss": 0.6222,
+      "r_loss": 0.0,
+      "step": 1340,
+      "steps": 1340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6922714913485306e-05,
+      "loss": 0.6213,
+      "r_loss": 0.0,
+      "step": 1350,
+      "steps": 1350
+    },
+    {
+      "epoch": 0.181,
+      "learning_rate": 1.689526366382862e-05,
+      "loss": 0.6278,
+      "r_loss": 0.0,
+      "step": 1360,
+      "steps": 1360
+    },
+    {
+      "epoch": 0.182,
+      "learning_rate": 1.6867812414171932e-05,
+      "loss": 0.615,
+      "r_loss": 0.0,
+      "step": 1370,
+      "steps": 1370
+    },
+    {
+      "epoch": 0.184,
+      "learning_rate": 1.6840361164515244e-05,
+      "loss": 0.6203,
+      "r_loss": 0.0,
+      "step": 1380,
+      "steps": 1380
+    },
+    {
+      "epoch": 0.185,
+      "learning_rate": 1.6812909914858556e-05,
+      "loss": 0.6144,
+      "r_loss": 0.0,
+      "step": 1390,
+      "steps": 1390
+    },
+    {
+      "epoch": 0.186,
+      "learning_rate": 1.6785458665201867e-05,
+      "loss": 0.6156,
+      "r_loss": 0.0,
+      "step": 1400,
+      "steps": 1400
+    },
+    {
+      "epoch": 0.188,
+      "learning_rate": 1.6758007415545182e-05,
+      "loss": 0.6251,
+      "r_loss": 0.0,
+      "step": 1410,
+      "steps": 1410
+    },
+    {
+      "epoch": 0.189,
+      "learning_rate": 1.6730556165888494e-05,
+      "loss": 0.6285,
+      "r_loss": 0.0,
+      "step": 1420,
+      "steps": 1420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6703104916231806e-05,
+      "loss": 0.6205,
+      "r_loss": 0.0,
+      "step": 1430,
+      "steps": 1430
+    },
+    {
+      "epoch": 0.192,
+      "learning_rate": 1.667565366657512e-05,
+      "loss": 0.6238,
+      "r_loss": 0.0,
+      "step": 1440,
+      "steps": 1440
+    },
+    {
+      "epoch": 0.193,
+      "learning_rate": 1.6648202416918432e-05,
+      "loss": 0.6257,
+      "r_loss": 0.0,
+      "step": 1450,
+      "steps": 1450
+    },
+    {
+      "epoch": 0.194,
+      "learning_rate": 1.6620751167261744e-05,
+      "loss": 0.6261,
+      "r_loss": 0.0,
+      "step": 1460,
+      "steps": 1460
+    },
+    {
+      "epoch": 0.196,
+      "learning_rate": 1.6593299917605056e-05,
+      "loss": 0.6316,
+      "r_loss": 0.0,
+      "step": 1470,
+      "steps": 1470
+    },
+    {
+      "epoch": 0.197,
+      "learning_rate": 1.6565848667948367e-05,
+      "loss": 0.6324,
+      "r_loss": 0.0,
+      "step": 1480,
+      "steps": 1480
+    },
+    {
+      "epoch": 0.198,
+      "learning_rate": 1.653839741829168e-05,
+      "loss": 0.6045,
+      "r_loss": 0.0,
+      "step": 1490,
+      "steps": 1490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6510946168634994e-05,
+      "loss": 0.6287,
+      "r_loss": 0.0,
+      "step": 1500,
+      "steps": 1500
+    },
+    {
+      "epoch": 0.201,
+      "learning_rate": 1.6483494918978305e-05,
+      "loss": 0.6268,
+      "r_loss": 0.0,
+      "step": 1510,
+      "steps": 1510
+    },
+    {
+      "epoch": 0.202,
+      "learning_rate": 1.6456043669321617e-05,
+      "loss": 0.6218,
+      "r_loss": 0.0,
+      "step": 1520,
+      "steps": 1520
+    },
+    {
+      "epoch": 0.204,
+      "learning_rate": 1.642859241966493e-05,
+      "loss": 0.6237,
+      "r_loss": 0.0,
+      "step": 1530,
+      "steps": 1530
+    },
+    {
+      "epoch": 0.205,
+      "learning_rate": 1.640114117000824e-05,
+      "loss": 0.6134,
+      "r_loss": 0.0,
+      "step": 1540,
+      "steps": 1540
+    },
+    {
+      "epoch": 0.206,
+      "learning_rate": 1.6373689920351552e-05,
+      "loss": 0.6046,
+      "r_loss": 0.0,
+      "step": 1550,
+      "steps": 1550
+    },
+    {
+      "epoch": 0.208,
+      "learning_rate": 1.6346238670694867e-05,
+      "loss": 0.612,
+      "r_loss": 0.0,
+      "step": 1560,
+      "steps": 1560
+    },
+    {
+      "epoch": 0.209,
+      "learning_rate": 1.631878742103818e-05,
+      "loss": 0.6143,
+      "r_loss": 0.0,
+      "step": 1570,
+      "steps": 1570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.629133617138149e-05,
+      "loss": 0.6069,
+      "r_loss": 0.0,
+      "step": 1580,
+      "steps": 1580
+    },
+    {
+      "epoch": 0.212,
+      "learning_rate": 1.6263884921724802e-05,
+      "loss": 0.6014,
+      "r_loss": 0.0,
+      "step": 1590,
+      "steps": 1590
+    },
+    {
+      "epoch": 0.213,
+      "learning_rate": 1.6236433672068114e-05,
+      "loss": 0.6092,
+      "r_loss": 0.0,
+      "step": 1600,
+      "steps": 1600
+    },
+    {
+      "epoch": 0.214,
+      "learning_rate": 1.620898242241143e-05,
+      "loss": 0.6284,
+      "r_loss": 0.0,
+      "step": 1610,
+      "steps": 1610
+    },
+    {
+      "epoch": 0.216,
+      "learning_rate": 1.618153117275474e-05,
+      "loss": 0.6235,
+      "r_loss": 0.0,
+      "step": 1620,
+      "steps": 1620
+    },
+    {
+      "epoch": 0.217,
+      "learning_rate": 1.6154079923098052e-05,
+      "loss": 0.6188,
+      "r_loss": 0.0,
+      "step": 1630,
+      "steps": 1630
+    },
+    {
+      "epoch": 0.218,
+      "learning_rate": 1.6126628673441363e-05,
+      "loss": 0.6202,
+      "r_loss": 0.0,
+      "step": 1640,
+      "steps": 1640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6099177423784675e-05,
+      "loss": 0.6147,
+      "r_loss": 0.0,
+      "step": 1650,
+      "steps": 1650
+    },
+    {
+      "epoch": 0.221,
+      "learning_rate": 1.6071726174127987e-05,
+      "loss": 0.6144,
+      "r_loss": 0.0,
+      "step": 1660,
+      "steps": 1660
+    },
+    {
+      "epoch": 0.222,
+      "learning_rate": 1.6044274924471302e-05,
+      "loss": 0.6188,
+      "r_loss": 0.0,
+      "step": 1670,
+      "steps": 1670
+    },
+    {
+      "epoch": 0.224,
+      "learning_rate": 1.6016823674814613e-05,
+      "loss": 0.611,
+      "r_loss": 0.0,
+      "step": 1680,
+      "steps": 1680
+    },
+    {
+      "epoch": 0.225,
+      "learning_rate": 1.5989372425157925e-05,
+      "loss": 0.6108,
+      "r_loss": 0.0,
+      "step": 1690,
+      "steps": 1690
+    },
+    {
+      "epoch": 0.226,
+      "learning_rate": 1.5961921175501237e-05,
+      "loss": 0.6087,
+      "r_loss": 0.0,
+      "step": 1700,
+      "steps": 1700
+    },
+    {
+      "epoch": 0.228,
+      "learning_rate": 1.593446992584455e-05,
+      "loss": 0.609,
+      "r_loss": 0.0,
+      "step": 1710,
+      "steps": 1710
+    },
+    {
+      "epoch": 0.229,
+      "learning_rate": 1.590701867618786e-05,
+      "loss": 0.6164,
+      "r_loss": 0.0,
+      "step": 1720,
+      "steps": 1720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5879567426531175e-05,
+      "loss": 0.6118,
+      "r_loss": 0.0,
+      "step": 1730,
+      "steps": 1730
+    },
+    {
+      "epoch": 0.232,
+      "learning_rate": 1.5852116176874487e-05,
+      "loss": 0.603,
+      "r_loss": 0.0,
+      "step": 1740,
+      "steps": 1740
+    },
+    {
+      "epoch": 0.233,
+      "learning_rate": 1.5824664927217798e-05,
+      "loss": 0.6195,
+      "r_loss": 0.0,
+      "step": 1750,
+      "steps": 1750
+    },
+    {
+      "epoch": 0.234,
+      "learning_rate": 1.5797213677561113e-05,
+      "loss": 0.6108,
+      "r_loss": 0.0,
+      "step": 1760,
+      "steps": 1760
+    },
+    {
+      "epoch": 0.236,
+      "learning_rate": 1.576976242790442e-05,
+      "loss": 0.6217,
+      "r_loss": 0.0,
+      "step": 1770,
+      "steps": 1770
+    },
+    {
+      "epoch": 0.237,
+      "learning_rate": 1.5742311178247737e-05,
+      "loss": 0.6152,
+      "r_loss": 0.0,
+      "step": 1780,
+      "steps": 1780
+    },
+    {
+      "epoch": 0.238,
+      "learning_rate": 1.5714859928591048e-05,
+      "loss": 0.6162,
+      "r_loss": 0.0,
+      "step": 1790,
+      "steps": 1790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.568740867893436e-05,
+      "loss": 0.6152,
+      "r_loss": 0.0,
+      "step": 1800,
+      "steps": 1800
+    },
+    {
+      "epoch": 0.241,
+      "learning_rate": 1.5659957429277675e-05,
+      "loss": 0.6155,
+      "r_loss": 0.0,
+      "step": 1810,
+      "steps": 1810
+    },
+    {
+      "epoch": 0.242,
+      "learning_rate": 1.5632506179620986e-05,
+      "loss": 0.6119,
+      "r_loss": 0.0,
+      "step": 1820,
+      "steps": 1820
+    },
+    {
+      "epoch": 0.244,
+      "learning_rate": 1.5605054929964298e-05,
+      "loss": 0.601,
+      "r_loss": 0.0,
+      "step": 1830,
+      "steps": 1830
+    },
+    {
+      "epoch": 0.245,
+      "learning_rate": 1.557760368030761e-05,
+      "loss": 0.6039,
+      "r_loss": 0.0,
+      "step": 1840,
+      "steps": 1840
+    },
+    {
+      "epoch": 0.246,
+      "learning_rate": 1.555015243065092e-05,
+      "loss": 0.6093,
+      "r_loss": 0.0,
+      "step": 1850,
+      "steps": 1850
+    },
+    {
+      "epoch": 0.248,
+      "learning_rate": 1.5522701180994233e-05,
+      "loss": 0.6041,
+      "r_loss": 0.0,
+      "step": 1860,
+      "steps": 1860
+    },
+    {
+      "epoch": 0.249,
+      "learning_rate": 1.5495249931337548e-05,
+      "loss": 0.6093,
+      "r_loss": 0.0,
+      "step": 1870,
+      "steps": 1870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.546779868168086e-05,
+      "loss": 0.6155,
+      "r_loss": 0.0,
+      "step": 1880,
+      "steps": 1880
+    },
+    {
+      "epoch": 0.252,
+      "learning_rate": 1.544034743202417e-05,
+      "loss": 0.6101,
+      "r_loss": 0.0,
+      "step": 1890,
+      "steps": 1890
+    },
+    {
+      "epoch": 0.253,
+      "learning_rate": 1.5412896182367483e-05,
+      "loss": 0.6089,
+      "r_loss": 0.0,
+      "step": 1900,
+      "steps": 1900
+    },
+    {
+      "epoch": 0.254,
+      "learning_rate": 1.5385444932710795e-05,
+      "loss": 0.6113,
+      "r_loss": 0.0,
+      "step": 1910,
+      "steps": 1910
+    },
+    {
+      "epoch": 0.256,
+      "learning_rate": 1.5357993683054106e-05,
+      "loss": 0.6082,
+      "r_loss": 0.0,
+      "step": 1920,
+      "steps": 1920
+    },
+    {
+      "epoch": 0.257,
+      "learning_rate": 1.533054243339742e-05,
+      "loss": 0.6079,
+      "r_loss": 0.0,
+      "step": 1930,
+      "steps": 1930
+    },
+    {
+      "epoch": 0.258,
+      "learning_rate": 1.5303091183740733e-05,
+      "loss": 0.6023,
+      "r_loss": 0.0,
+      "step": 1940,
+      "steps": 1940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5275639934084044e-05,
+      "loss": 0.613,
+      "r_loss": 0.0,
+      "step": 1950,
+      "steps": 1950
+    },
+    {
+      "epoch": 0.261,
+      "learning_rate": 1.5248188684427356e-05,
+      "loss": 0.611,
+      "r_loss": 0.0,
+      "step": 1960,
+      "steps": 1960
+    },
+    {
+      "epoch": 0.262,
+      "learning_rate": 1.5220737434770668e-05,
+      "loss": 0.621,
+      "r_loss": 0.0,
+      "step": 1970,
+      "steps": 1970
+    },
+    {
+      "epoch": 0.264,
+      "learning_rate": 1.5193286185113981e-05,
+      "loss": 0.6062,
+      "r_loss": 0.0,
+      "step": 1980,
+      "steps": 1980
+    },
+    {
+      "epoch": 0.265,
+      "learning_rate": 1.5165834935457293e-05,
+      "loss": 0.6073,
+      "r_loss": 0.0,
+      "step": 1990,
+      "steps": 1990
+    },
+    {
+      "epoch": 0.266,
+      "learning_rate": 1.5138383685800604e-05,
+      "loss": 0.6124,
+      "r_loss": 0.0,
+      "step": 2000,
+      "steps": 2000
+    },
+    {
+      "epoch": 0.268,
+      "learning_rate": 1.5110932436143918e-05,
+      "loss": 0.5875,
+      "r_loss": 0.0,
+      "step": 2010,
+      "steps": 2010
+    },
+    {
+      "epoch": 0.269,
+      "learning_rate": 1.508348118648723e-05,
+      "loss": 0.6045,
+      "r_loss": 0.0,
+      "step": 2020,
+      "steps": 2020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5056029936830541e-05,
+      "loss": 0.6004,
+      "r_loss": 0.0,
+      "step": 2030,
+      "steps": 2030
+    },
+    {
+      "epoch": 0.272,
+      "learning_rate": 1.5028578687173854e-05,
+      "loss": 0.5976,
+      "r_loss": 0.0,
+      "step": 2040,
+      "steps": 2040
+    },
+    {
+      "epoch": 0.273,
+      "learning_rate": 1.5001127437517166e-05,
+      "loss": 0.6102,
+      "r_loss": 0.0,
+      "step": 2050,
+      "steps": 2050
+    },
+    {
+      "epoch": 0.274,
+      "learning_rate": 1.4973676187860477e-05,
+      "loss": 0.6055,
+      "r_loss": 0.0,
+      "step": 2060,
+      "steps": 2060
+    },
+    {
+      "epoch": 0.276,
+      "learning_rate": 1.494622493820379e-05,
+      "loss": 0.6121,
+      "r_loss": 0.0,
+      "step": 2070,
+      "steps": 2070
+    },
+    {
+      "epoch": 0.277,
+      "learning_rate": 1.4918773688547102e-05,
+      "loss": 0.5986,
+      "r_loss": 0.0,
+      "step": 2080,
+      "steps": 2080
+    },
+    {
+      "epoch": 0.278,
+      "learning_rate": 1.4891322438890414e-05,
+      "loss": 0.5981,
+      "r_loss": 0.0,
+      "step": 2090,
+      "steps": 2090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4863871189233727e-05,
+      "loss": 0.6056,
+      "r_loss": 0.0,
+      "step": 2100,
+      "steps": 2100
+    },
+    {
+      "epoch": 0.281,
+      "learning_rate": 1.4836419939577039e-05,
+      "loss": 0.6012,
+      "r_loss": 0.0,
+      "step": 2110,
+      "steps": 2110
+    },
+    {
+      "epoch": 0.282,
+      "learning_rate": 1.480896868992035e-05,
+      "loss": 0.5978,
+      "r_loss": 0.0,
+      "step": 2120,
+      "steps": 2120
+    },
+    {
+      "epoch": 0.284,
+      "learning_rate": 1.4781517440263664e-05,
+      "loss": 0.6054,
+      "r_loss": 0.0,
+      "step": 2130,
+      "steps": 2130
+    },
+    {
+      "epoch": 0.285,
+      "learning_rate": 1.4754066190606976e-05,
+      "loss": 0.5973,
+      "r_loss": 0.0,
+      "step": 2140,
+      "steps": 2140
+    },
+    {
+      "epoch": 0.286,
+      "learning_rate": 1.4726614940950289e-05,
+      "loss": 0.5924,
+      "r_loss": 0.0,
+      "step": 2150,
+      "steps": 2150
+    },
+    {
+      "epoch": 0.288,
+      "learning_rate": 1.46991636912936e-05,
+      "loss": 0.608,
+      "r_loss": 0.0,
+      "step": 2160,
+      "steps": 2160
+    },
+    {
+      "epoch": 0.289,
+      "learning_rate": 1.4671712441636912e-05,
+      "loss": 0.613,
+      "r_loss": 0.0,
+      "step": 2170,
+      "steps": 2170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4644261191980226e-05,
+      "loss": 0.5915,
+      "r_loss": 0.0,
+      "step": 2180,
+      "steps": 2180
+    },
+    {
+      "epoch": 0.292,
+      "learning_rate": 1.4616809942323537e-05,
+      "loss": 0.6014,
+      "r_loss": 0.0,
+      "step": 2190,
+      "steps": 2190
+    },
+    {
+      "epoch": 0.293,
+      "learning_rate": 1.4589358692666849e-05,
+      "loss": 0.5866,
+      "r_loss": 0.0,
+      "step": 2200,
+      "steps": 2200
+    },
+    {
+      "epoch": 0.294,
+      "learning_rate": 1.4561907443010162e-05,
+      "loss": 0.6016,
+      "r_loss": 0.0,
+      "step": 2210,
+      "steps": 2210
+    },
+    {
+      "epoch": 0.296,
+      "learning_rate": 1.4534456193353474e-05,
+      "loss": 0.5869,
+      "r_loss": 0.0,
+      "step": 2220,
+      "steps": 2220
+    },
+    {
+      "epoch": 0.297,
+      "learning_rate": 1.4507004943696785e-05,
+      "loss": 0.5988,
+      "r_loss": 0.0,
+      "step": 2230,
+      "steps": 2230
+    },
+    {
+      "epoch": 0.298,
+      "learning_rate": 1.4479553694040099e-05,
+      "loss": 0.6068,
+      "r_loss": 0.0,
+      "step": 2240,
+      "steps": 2240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.445210244438341e-05,
+      "loss": 0.6059,
+      "r_loss": 0.0,
+      "step": 2250,
+      "steps": 2250
+    },
+    {
+      "epoch": 0.301,
+      "learning_rate": 1.4424651194726722e-05,
+      "loss": 0.5972,
+      "r_loss": 0.0,
+      "step": 2260,
+      "steps": 2260
+    },
+    {
+      "epoch": 0.302,
+      "learning_rate": 1.4397199945070037e-05,
+      "loss": 0.5969,
+      "r_loss": 0.0,
+      "step": 2270,
+      "steps": 2270
+    },
+    {
+      "epoch": 0.304,
+      "learning_rate": 1.4369748695413347e-05,
+      "loss": 0.5935,
+      "r_loss": 0.0,
+      "step": 2280,
+      "steps": 2280
+    },
+    {
+      "epoch": 0.305,
+      "learning_rate": 1.4342297445756659e-05,
+      "loss": 0.5932,
+      "r_loss": 0.0,
+      "step": 2290,
+      "steps": 2290
+    },
+    {
+      "epoch": 0.306,
+      "learning_rate": 1.4314846196099974e-05,
+      "loss": 0.6008,
+      "r_loss": 0.0,
+      "step": 2300,
+      "steps": 2300
+    },
+    {
+      "epoch": 0.308,
+      "learning_rate": 1.4287394946443284e-05,
+      "loss": 0.5898,
+      "r_loss": 0.0,
+      "step": 2310,
+      "steps": 2310
+    },
+    {
+      "epoch": 0.309,
+      "learning_rate": 1.4259943696786599e-05,
+      "loss": 0.6052,
+      "r_loss": 0.0,
+      "step": 2320,
+      "steps": 2320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.423249244712991e-05,
+      "loss": 0.5929,
+      "r_loss": 0.0,
+      "step": 2330,
+      "steps": 2330
+    },
+    {
+      "epoch": 0.312,
+      "learning_rate": 1.4205041197473222e-05,
+      "loss": 0.5946,
+      "r_loss": 0.0,
+      "step": 2340,
+      "steps": 2340
+    },
+    {
+      "epoch": 0.313,
+      "learning_rate": 1.4177589947816535e-05,
+      "loss": 0.5973,
+      "r_loss": 0.0,
+      "step": 2350,
+      "steps": 2350
+    },
+    {
+      "epoch": 0.314,
+      "learning_rate": 1.4150138698159847e-05,
+      "loss": 0.6,
+      "r_loss": 0.0,
+      "step": 2360,
+      "steps": 2360
+    },
+    {
+      "epoch": 0.316,
+      "learning_rate": 1.4122687448503158e-05,
+      "loss": 0.582,
+      "r_loss": 0.0,
+      "step": 2370,
+      "steps": 2370
+    },
+    {
+      "epoch": 0.317,
+      "learning_rate": 1.4095236198846472e-05,
+      "loss": 0.5976,
+      "r_loss": 0.0,
+      "step": 2380,
+      "steps": 2380
+    },
+    {
+      "epoch": 0.318,
+      "learning_rate": 1.4067784949189783e-05,
+      "loss": 0.5907,
+      "r_loss": 0.0,
+      "step": 2390,
+      "steps": 2390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4040333699533095e-05,
+      "loss": 0.6032,
+      "r_loss": 0.0,
+      "step": 2400,
+      "steps": 2400
+    },
+    {
+      "epoch": 0.321,
+      "learning_rate": 1.4012882449876408e-05,
+      "loss": 0.6027,
+      "r_loss": 0.0,
+      "step": 2410,
+      "steps": 2410
+    },
+    {
+      "epoch": 0.322,
+      "learning_rate": 1.398543120021972e-05,
+      "loss": 0.5967,
+      "r_loss": 0.0,
+      "step": 2420,
+      "steps": 2420
+    },
+    {
+      "epoch": 0.324,
+      "learning_rate": 1.3957979950563032e-05,
+      "loss": 0.5966,
+      "r_loss": 0.0,
+      "step": 2430,
+      "steps": 2430
+    },
+    {
+      "epoch": 0.325,
+      "learning_rate": 1.3930528700906345e-05,
+      "loss": 0.5963,
+      "r_loss": 0.0,
+      "step": 2440,
+      "steps": 2440
+    },
+    {
+      "epoch": 0.326,
+      "learning_rate": 1.3903077451249657e-05,
+      "loss": 0.5932,
+      "r_loss": 0.0,
+      "step": 2450,
+      "steps": 2450
+    },
+    {
+      "epoch": 0.328,
+      "learning_rate": 1.3875626201592968e-05,
+      "loss": 0.5934,
+      "r_loss": 0.0,
+      "step": 2460,
+      "steps": 2460
+    },
+    {
+      "epoch": 0.329,
+      "learning_rate": 1.3848174951936282e-05,
+      "loss": 0.588,
+      "r_loss": 0.0,
+      "step": 2470,
+      "steps": 2470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3820723702279593e-05,
+      "loss": 0.593,
+      "r_loss": 0.0,
+      "step": 2480,
+      "steps": 2480
+    },
+    {
+      "epoch": 0.332,
+      "learning_rate": 1.3793272452622905e-05,
+      "loss": 0.5936,
+      "r_loss": 0.0,
+      "step": 2490,
+      "steps": 2490
+    },
+    {
+      "epoch": 0.333,
+      "learning_rate": 1.3765821202966218e-05,
+      "loss": 0.5741,
+      "r_loss": 0.0,
+      "step": 2500,
+      "steps": 2500
+    },
+    {
+      "epoch": 0.334,
+      "learning_rate": 1.373836995330953e-05,
+      "loss": 0.6024,
+      "r_loss": 0.0,
+      "step": 2510,
+      "steps": 2510
+    },
+    {
+      "epoch": 0.336,
+      "learning_rate": 1.3710918703652843e-05,
+      "loss": 0.5808,
+      "r_loss": 0.0,
+      "step": 2520,
+      "steps": 2520
+    },
+    {
+      "epoch": 0.337,
+      "learning_rate": 1.3683467453996155e-05,
+      "loss": 0.6057,
+      "r_loss": 0.0,
+      "step": 2530,
+      "steps": 2530
+    },
+    {
+      "epoch": 0.338,
+      "learning_rate": 1.3656016204339466e-05,
+      "loss": 0.589,
+      "r_loss": 0.0,
+      "step": 2540,
+      "steps": 2540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.362856495468278e-05,
+      "loss": 0.6035,
+      "r_loss": 0.0,
+      "step": 2550,
+      "steps": 2550
+    },
+    {
+      "epoch": 0.341,
+      "learning_rate": 1.3601113705026091e-05,
+      "loss": 0.5855,
+      "r_loss": 0.0,
+      "step": 2560,
+      "steps": 2560
+    },
+    {
+      "epoch": 0.342,
+      "learning_rate": 1.3573662455369403e-05,
+      "loss": 0.582,
+      "r_loss": 0.0,
+      "step": 2570,
+      "steps": 2570
+    },
+    {
+      "epoch": 0.344,
+      "learning_rate": 1.3546211205712716e-05,
+      "loss": 0.5968,
+      "r_loss": 0.0,
+      "step": 2580,
+      "steps": 2580
+    },
+    {
+      "epoch": 0.345,
+      "learning_rate": 1.3518759956056028e-05,
+      "loss": 0.58,
+      "r_loss": 0.0,
+      "step": 2590,
+      "steps": 2590
+    },
+    {
+      "epoch": 0.346,
+      "learning_rate": 1.349130870639934e-05,
+      "loss": 0.5839,
+      "r_loss": 0.0,
+      "step": 2600,
+      "steps": 2600
+    },
+    {
+      "epoch": 0.348,
+      "learning_rate": 1.3463857456742653e-05,
+      "loss": 0.5938,
+      "r_loss": 0.0,
+      "step": 2610,
+      "steps": 2610
+    },
+    {
+      "epoch": 0.349,
+      "learning_rate": 1.3441896457017303e-05,
+      "loss": 0.5965,
+      "r_loss": 0.0,
+      "step": 2620,
+      "steps": 2620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3414445207360614e-05,
+      "loss": 0.5904,
+      "r_loss": 0.0,
+      "step": 2630,
+      "steps": 2630
+    },
+    {
+      "epoch": 0.352,
+      "learning_rate": 1.3386993957703927e-05,
+      "loss": 0.5884,
+      "r_loss": 0.0,
+      "step": 2640,
+      "steps": 2640
+    },
+    {
+      "epoch": 0.353,
+      "learning_rate": 1.3359542708047239e-05,
+      "loss": 0.6003,
+      "r_loss": 0.0,
+      "step": 2650,
+      "steps": 2650
+    },
+    {
+      "epoch": 0.354,
+      "learning_rate": 1.333209145839055e-05,
+      "loss": 0.6024,
+      "r_loss": 0.0,
+      "step": 2660,
+      "steps": 2660
+    },
+    {
+      "epoch": 0.356,
+      "learning_rate": 1.3304640208733864e-05,
+      "loss": 0.5919,
+      "r_loss": 0.0,
+      "step": 2670,
+      "steps": 2670
+    },
+    {
+      "epoch": 0.357,
+      "learning_rate": 1.3277188959077176e-05,
+      "loss": 0.5792,
+      "r_loss": 0.0,
+      "step": 2680,
+      "steps": 2680
+    },
+    {
+      "epoch": 0.358,
+      "learning_rate": 1.3249737709420487e-05,
+      "loss": 0.5788,
+      "r_loss": 0.0,
+      "step": 2690,
+      "steps": 2690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3222286459763802e-05,
+      "loss": 0.5836,
+      "r_loss": 0.0,
+      "step": 2700,
+      "steps": 2700
+    },
+    {
+      "epoch": 0.361,
+      "learning_rate": 1.3194835210107112e-05,
+      "loss": 0.5938,
+      "r_loss": 0.0,
+      "step": 2710,
+      "steps": 2710
+    },
+    {
+      "epoch": 0.362,
+      "learning_rate": 1.3167383960450427e-05,
+      "loss": 0.5865,
+      "r_loss": 0.0,
+      "step": 2720,
+      "steps": 2720
+    },
+    {
+      "epoch": 0.364,
+      "learning_rate": 1.3139932710793739e-05,
+      "loss": 0.585,
+      "r_loss": 0.0,
+      "step": 2730,
+      "steps": 2730
+    },
+    {
+      "epoch": 0.365,
+      "learning_rate": 1.311248146113705e-05,
+      "loss": 0.5785,
+      "r_loss": 0.0,
+      "step": 2740,
+      "steps": 2740
+    },
+    {
+      "epoch": 0.366,
+      "learning_rate": 1.3085030211480364e-05,
+      "loss": 0.5884,
+      "r_loss": 0.0,
+      "step": 2750,
+      "steps": 2750
+    },
+    {
+      "epoch": 0.368,
+      "learning_rate": 1.3057578961823676e-05,
+      "loss": 0.5992,
+      "r_loss": 0.0,
+      "step": 2760,
+      "steps": 2760
+    },
+    {
+      "epoch": 0.369,
+      "learning_rate": 1.3030127712166987e-05,
+      "loss": 0.5862,
+      "r_loss": 0.0,
+      "step": 2770,
+      "steps": 2770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.30026764625103e-05,
+      "loss": 0.5929,
+      "r_loss": 0.0,
+      "step": 2780,
+      "steps": 2780
+    },
+    {
+      "epoch": 0.372,
+      "learning_rate": 1.2975225212853612e-05,
+      "loss": 0.5876,
+      "r_loss": 0.0,
+      "step": 2790,
+      "steps": 2790
+    },
+    {
+      "epoch": 0.373,
+      "learning_rate": 1.2947773963196924e-05,
+      "loss": 0.5741,
+      "r_loss": 0.0,
+      "step": 2800,
+      "steps": 2800
+    },
+    {
+      "epoch": 0.374,
+      "learning_rate": 1.2920322713540237e-05,
+      "loss": 0.5762,
+      "r_loss": 0.0,
+      "step": 2810,
+      "steps": 2810
+    },
+    {
+      "epoch": 0.376,
+      "learning_rate": 1.2892871463883549e-05,
+      "loss": 0.5776,
+      "r_loss": 0.0,
+      "step": 2820,
+      "steps": 2820
+    },
+    {
+      "epoch": 0.377,
+      "learning_rate": 1.286542021422686e-05,
+      "loss": 0.5848,
+      "r_loss": 0.0,
+      "step": 2830,
+      "steps": 2830
+    },
+    {
+      "epoch": 0.378,
+      "learning_rate": 1.2837968964570174e-05,
+      "loss": 0.575,
+      "r_loss": 0.0,
+      "step": 2840,
+      "steps": 2840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2810517714913485e-05,
+      "loss": 0.5819,
+      "r_loss": 0.0,
+      "step": 2850,
+      "steps": 2850
+    },
+    {
+      "epoch": 0.381,
+      "learning_rate": 1.2783066465256797e-05,
+      "loss": 0.5886,
+      "r_loss": 0.0,
+      "step": 2860,
+      "steps": 2860
+    },
+    {
+      "epoch": 0.382,
+      "learning_rate": 1.275561521560011e-05,
+      "loss": 0.5836,
+      "r_loss": 0.0,
+      "step": 2870,
+      "steps": 2870
+    },
+    {
+      "epoch": 0.384,
+      "learning_rate": 1.2728163965943422e-05,
+      "loss": 0.5787,
+      "r_loss": 0.0,
+      "step": 2880,
+      "steps": 2880
+    },
+    {
+      "epoch": 0.385,
+      "learning_rate": 1.2700712716286734e-05,
+      "loss": 0.582,
+      "r_loss": 0.0,
+      "step": 2890,
+      "steps": 2890
+    },
+    {
+      "epoch": 0.386,
+      "learning_rate": 1.2673261466630047e-05,
+      "loss": 0.571,
+      "r_loss": 0.0,
+      "step": 2900,
+      "steps": 2900
+    },
+    {
+      "epoch": 0.388,
+      "learning_rate": 1.2645810216973359e-05,
+      "loss": 0.5947,
+      "r_loss": 0.0,
+      "step": 2910,
+      "steps": 2910
+    },
+    {
+      "epoch": 0.389,
+      "learning_rate": 1.2618358967316672e-05,
+      "loss": 0.5769,
+      "r_loss": 0.0,
+      "step": 2920,
+      "steps": 2920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2590907717659983e-05,
+      "loss": 0.5744,
+      "r_loss": 0.0,
+      "step": 2930,
+      "steps": 2930
+    },
+    {
+      "epoch": 0.392,
+      "learning_rate": 1.2563456468003295e-05,
+      "loss": 0.5913,
+      "r_loss": 0.0,
+      "step": 2940,
+      "steps": 2940
+    },
+    {
+      "epoch": 0.393,
+      "learning_rate": 1.2536005218346608e-05,
+      "loss": 0.5904,
+      "r_loss": 0.0,
+      "step": 2950,
+      "steps": 2950
+    },
+    {
+      "epoch": 0.394,
+      "learning_rate": 1.250855396868992e-05,
+      "loss": 0.5814,
+      "r_loss": 0.0,
+      "step": 2960,
+      "steps": 2960
+    },
+    {
+      "epoch": 0.396,
+      "learning_rate": 1.2481102719033232e-05,
+      "loss": 0.5824,
+      "r_loss": 0.0,
+      "step": 2970,
+      "steps": 2970
+    },
+    {
+      "epoch": 0.397,
+      "learning_rate": 1.2453651469376545e-05,
+      "loss": 0.5823,
+      "r_loss": 0.0,
+      "step": 2980,
+      "steps": 2980
+    },
+    {
+      "epoch": 0.398,
+      "learning_rate": 1.2426200219719857e-05,
+      "loss": 0.586,
+      "r_loss": 0.0,
+      "step": 2990,
+      "steps": 2990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2398748970063168e-05,
+      "loss": 0.5777,
+      "r_loss": 0.0,
+      "step": 3000,
+      "steps": 3000
+    },
+    {
+      "epoch": 0.401,
+      "learning_rate": 1.2371297720406482e-05,
+      "loss": 0.5626,
+      "r_loss": 0.0,
+      "step": 3010,
+      "steps": 3010
+    },
+    {
+      "epoch": 0.402,
+      "learning_rate": 1.2343846470749793e-05,
+      "loss": 0.5889,
+      "r_loss": 0.0,
+      "step": 3020,
+      "steps": 3020
+    },
+    {
+      "epoch": 0.404,
+      "learning_rate": 1.2316395221093105e-05,
+      "loss": 0.5844,
+      "r_loss": 0.0,
+      "step": 3030,
+      "steps": 3030
+    },
+    {
+      "epoch": 0.405,
+      "learning_rate": 1.228894397143642e-05,
+      "loss": 0.5813,
+      "r_loss": 0.0,
+      "step": 3040,
+      "steps": 3040
+    },
+    {
+      "epoch": 0.406,
+      "learning_rate": 1.226149272177973e-05,
+      "loss": 0.5916,
+      "r_loss": 0.0,
+      "step": 3050,
+      "steps": 3050
+    },
+    {
+      "epoch": 0.408,
+      "learning_rate": 1.2234041472123041e-05,
+      "loss": 0.5791,
+      "r_loss": 0.0,
+      "step": 3060,
+      "steps": 3060
+    },
+    {
+      "epoch": 0.409,
+      "learning_rate": 1.2206590222466356e-05,
+      "loss": 0.5968,
+      "r_loss": 0.0,
+      "step": 3070,
+      "steps": 3070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2179138972809666e-05,
+      "loss": 0.5795,
+      "r_loss": 0.0,
+      "step": 3080,
+      "steps": 3080
+    },
+    {
+      "epoch": 0.412,
+      "learning_rate": 1.2151687723152981e-05,
+      "loss": 0.5632,
+      "r_loss": 0.0,
+      "step": 3090,
+      "steps": 3090
+    },
+    {
+      "epoch": 0.413,
+      "learning_rate": 1.2124236473496293e-05,
+      "loss": 0.5717,
+      "r_loss": 0.0,
+      "step": 3100,
+      "steps": 3100
+    },
+    {
+      "epoch": 0.414,
+      "learning_rate": 1.2096785223839605e-05,
+      "loss": 0.5803,
+      "r_loss": 0.0,
+      "step": 3110,
+      "steps": 3110
+    },
+    {
+      "epoch": 0.416,
+      "learning_rate": 1.2069333974182918e-05,
+      "loss": 0.585,
+      "r_loss": 0.0,
+      "step": 3120,
+      "steps": 3120
+    },
+    {
+      "epoch": 0.417,
+      "learning_rate": 1.204188272452623e-05,
+      "loss": 0.5777,
+      "r_loss": 0.0,
+      "step": 3130,
+      "steps": 3130
+    },
+    {
+      "epoch": 0.418,
+      "learning_rate": 1.2014431474869541e-05,
+      "loss": 0.5785,
+      "r_loss": 0.0,
+      "step": 3140,
+      "steps": 3140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1986980225212855e-05,
+      "loss": 0.5854,
+      "r_loss": 0.0,
+      "step": 3150,
+      "steps": 3150
+    },
+    {
+      "epoch": 0.421,
+      "learning_rate": 1.1959528975556166e-05,
+      "loss": 0.5809,
+      "r_loss": 0.0,
+      "step": 3160,
+      "steps": 3160
+    },
+    {
+      "epoch": 0.422,
+      "learning_rate": 1.1932077725899478e-05,
+      "loss": 0.5864,
+      "r_loss": 0.0,
+      "step": 3170,
+      "steps": 3170
+    },
+    {
+      "epoch": 0.424,
+      "learning_rate": 1.1904626476242791e-05,
+      "loss": 0.5707,
+      "r_loss": 0.0,
+      "step": 3180,
+      "steps": 3180
+    },
+    {
+      "epoch": 0.425,
+      "learning_rate": 1.1877175226586103e-05,
+      "loss": 0.5837,
+      "r_loss": 0.0,
+      "step": 3190,
+      "steps": 3190
+    },
+    {
+      "epoch": 0.426,
+      "learning_rate": 1.1849723976929414e-05,
+      "loss": 0.569,
+      "r_loss": 0.0,
+      "step": 3200,
+      "steps": 3200
+    },
+    {
+      "epoch": 0.428,
+      "learning_rate": 1.1822272727272728e-05,
+      "loss": 0.5645,
+      "r_loss": 0.0,
+      "step": 3210,
+      "steps": 3210
+    },
+    {
+      "epoch": 0.429,
+      "learning_rate": 1.179482147761604e-05,
+      "loss": 0.5774,
+      "r_loss": 0.0,
+      "step": 3220,
+      "steps": 3220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1767370227959351e-05,
+      "loss": 0.5816,
+      "r_loss": 0.0,
+      "step": 3230,
+      "steps": 3230
+    },
+    {
+      "epoch": 0.432,
+      "learning_rate": 1.1739918978302664e-05,
+      "loss": 0.5741,
+      "r_loss": 0.0,
+      "step": 3240,
+      "steps": 3240
+    },
+    {
+      "epoch": 0.433,
+      "learning_rate": 1.1712467728645976e-05,
+      "loss": 0.5811,
+      "r_loss": 0.0,
+      "step": 3250,
+      "steps": 3250
+    },
+    {
+      "epoch": 0.434,
+      "learning_rate": 1.1685016478989288e-05,
+      "loss": 0.5806,
+      "r_loss": 0.0,
+      "step": 3260,
+      "steps": 3260
+    },
+    {
+      "epoch": 0.436,
+      "learning_rate": 1.1657565229332601e-05,
+      "loss": 0.571,
+      "r_loss": 0.0,
+      "step": 3270,
+      "steps": 3270
+    },
+    {
+      "epoch": 0.437,
+      "learning_rate": 1.1630113979675913e-05,
+      "loss": 0.577,
+      "r_loss": 0.0,
+      "step": 3280,
+      "steps": 3280
+    },
+    {
+      "epoch": 0.438,
+      "learning_rate": 1.1602662730019226e-05,
+      "loss": 0.5806,
+      "r_loss": 0.0,
+      "step": 3290,
+      "steps": 3290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1575211480362538e-05,
+      "loss": 0.5725,
+      "r_loss": 0.0,
+      "step": 3300,
+      "steps": 3300
+    },
+    {
+      "epoch": 0.441,
+      "learning_rate": 1.154776023070585e-05,
+      "loss": 0.583,
+      "r_loss": 0.0,
+      "step": 3310,
+      "steps": 3310
+    },
+    {
+      "epoch": 0.442,
+      "learning_rate": 1.1520308981049163e-05,
+      "loss": 0.5607,
+      "r_loss": 0.0,
+      "step": 3320,
+      "steps": 3320
+    },
+    {
+      "epoch": 0.444,
+      "learning_rate": 1.1492857731392474e-05,
+      "loss": 0.5822,
+      "r_loss": 0.0,
+      "step": 3330,
+      "steps": 3330
+    },
+    {
+      "epoch": 0.445,
+      "learning_rate": 1.1465406481735786e-05,
+      "loss": 0.5712,
+      "r_loss": 0.0,
+      "step": 3340,
+      "steps": 3340
+    },
+    {
+      "epoch": 0.446,
+      "learning_rate": 1.1437955232079099e-05,
+      "loss": 0.5704,
+      "r_loss": 0.0,
+      "step": 3350,
+      "steps": 3350
+    },
+    {
+      "epoch": 0.448,
+      "learning_rate": 1.141050398242241e-05,
+      "loss": 0.5645,
+      "r_loss": 0.0,
+      "step": 3360,
+      "steps": 3360
+    },
+    {
+      "epoch": 0.449,
+      "learning_rate": 1.1383052732765722e-05,
+      "loss": 0.5697,
+      "r_loss": 0.0,
+      "step": 3370,
+      "steps": 3370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1355601483109036e-05,
+      "loss": 0.5835,
+      "r_loss": 0.0,
+      "step": 3380,
+      "steps": 3380
+    },
+    {
+      "epoch": 0.452,
+      "learning_rate": 1.1328150233452347e-05,
+      "loss": 0.5855,
+      "r_loss": 0.0,
+      "step": 3390,
+      "steps": 3390
+    },
+    {
+      "epoch": 0.453,
+      "learning_rate": 1.1300698983795659e-05,
+      "loss": 0.5644,
+      "r_loss": 0.0,
+      "step": 3400,
+      "steps": 3400
+    },
+    {
+      "epoch": 0.454,
+      "learning_rate": 1.1273247734138974e-05,
+      "loss": 0.5634,
+      "r_loss": 0.0,
+      "step": 3410,
+      "steps": 3410
+    },
+    {
+      "epoch": 0.456,
+      "learning_rate": 1.1245796484482284e-05,
+      "loss": 0.5747,
+      "r_loss": 0.0,
+      "step": 3420,
+      "steps": 3420
+    },
+    {
+      "epoch": 0.457,
+      "learning_rate": 1.1218345234825596e-05,
+      "loss": 0.5751,
+      "r_loss": 0.0,
+      "step": 3430,
+      "steps": 3430
+    },
+    {
+      "epoch": 0.458,
+      "learning_rate": 1.119089398516891e-05,
+      "loss": 0.5728,
+      "r_loss": 0.0,
+      "step": 3440,
+      "steps": 3440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1163442735512222e-05,
+      "loss": 0.5642,
+      "r_loss": 0.0,
+      "step": 3450,
+      "steps": 3450
+    },
+    {
+      "epoch": 0.461,
+      "learning_rate": 1.1135991485855536e-05,
+      "loss": 0.5725,
+      "r_loss": 0.0,
+      "step": 3460,
+      "steps": 3460
+    },
+    {
+      "epoch": 0.462,
+      "learning_rate": 1.1108540236198847e-05,
+      "loss": 0.5699,
+      "r_loss": 0.0,
+      "step": 3470,
+      "steps": 3470
+    },
+    {
+      "epoch": 0.464,
+      "learning_rate": 1.1081088986542159e-05,
+      "loss": 0.5726,
+      "r_loss": 0.0,
+      "step": 3480,
+      "steps": 3480
+    },
+    {
+      "epoch": 0.465,
+      "learning_rate": 1.1053637736885472e-05,
+      "loss": 0.5708,
+      "r_loss": 0.0,
+      "step": 3490,
+      "steps": 3490
+    },
+    {
+      "epoch": 0.466,
+      "learning_rate": 1.1026186487228784e-05,
+      "loss": 0.5739,
+      "r_loss": 0.0,
+      "step": 3500,
+      "steps": 3500
+    },
+    {
+      "epoch": 0.468,
+      "learning_rate": 1.0998735237572095e-05,
+      "loss": 0.5696,
+      "r_loss": 0.0,
+      "step": 3510,
+      "steps": 3510
+    },
+    {
+      "epoch": 0.469,
+      "learning_rate": 1.0971283987915409e-05,
+      "loss": 0.5792,
+      "r_loss": 0.0,
+      "step": 3520,
+      "steps": 3520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.094383273825872e-05,
+      "loss": 0.5677,
+      "r_loss": 0.0,
+      "step": 3530,
+      "steps": 3530
+    },
+    {
+      "epoch": 0.471,
+      "learning_rate": 1.0916381488602032e-05,
+      "loss": 0.5726,
+      "r_loss": 0.0,
+      "step": 3540,
+      "steps": 3540
+    },
+    {
+      "epoch": 0.473,
+      "learning_rate": 1.0888930238945345e-05,
+      "loss": 0.5643,
+      "r_loss": 0.0,
+      "step": 3550,
+      "steps": 3550
+    },
+    {
+      "epoch": 0.474,
+      "learning_rate": 1.0861478989288657e-05,
+      "loss": 0.5656,
+      "r_loss": 0.0,
+      "step": 3560,
+      "steps": 3560
+    },
+    {
+      "epoch": 0.475,
+      "learning_rate": 1.0834027739631969e-05,
+      "loss": 0.5645,
+      "r_loss": 0.0,
+      "step": 3570,
+      "steps": 3570
+    },
+    {
+      "epoch": 0.477,
+      "learning_rate": 1.0806576489975282e-05,
+      "loss": 0.5779,
+      "r_loss": 0.0,
+      "step": 3580,
+      "steps": 3580
+    },
+    {
+      "epoch": 0.478,
+      "learning_rate": 1.0779125240318594e-05,
+      "loss": 0.5722,
+      "r_loss": 0.0,
+      "step": 3590,
+      "steps": 3590
+    },
+    {
+      "epoch": 0.479,
+      "learning_rate": 1.0751673990661905e-05,
+      "loss": 0.5718,
+      "r_loss": 0.0,
+      "step": 3600,
+      "steps": 3600
+    },
+    {
+      "epoch": 0.481,
+      "learning_rate": 1.0724222741005219e-05,
+      "loss": 0.5662,
+      "r_loss": 0.0,
+      "step": 3610,
+      "steps": 3610
+    },
+    {
+      "epoch": 0.482,
+      "learning_rate": 1.0699516616314198e-05,
+      "loss": 0.561,
+      "r_loss": 0.0,
+      "step": 3620,
+      "steps": 3620
+    },
+    {
+      "epoch": 0.483,
+      "learning_rate": 1.067481049162318e-05,
+      "loss": 0.5697,
+      "r_loss": 0.0,
+      "step": 3630,
+      "steps": 3630
+    },
+    {
+      "epoch": 0.485,
+      "learning_rate": 1.0647359241966493e-05,
+      "loss": 0.5617,
+      "r_loss": 0.0,
+      "step": 3640,
+      "steps": 3640
+    },
+    {
+      "epoch": 0.486,
+      "learning_rate": 1.0619907992309805e-05,
+      "loss": 0.5542,
+      "r_loss": 0.0,
+      "step": 3650,
+      "steps": 3650
+    },
+    {
+      "epoch": 0.487,
+      "learning_rate": 1.0592456742653116e-05,
+      "loss": 0.5873,
+      "r_loss": 0.0,
+      "step": 3660,
+      "steps": 3660
+    },
+    {
+      "epoch": 0.489,
+      "learning_rate": 1.056500549299643e-05,
+      "loss": 0.5715,
+      "r_loss": 0.0,
+      "step": 3670,
+      "steps": 3670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0537554243339741e-05,
+      "loss": 0.5665,
+      "r_loss": 0.0,
+      "step": 3680,
+      "steps": 3680
+    },
+    {
+      "epoch": 0.491,
+      "learning_rate": 1.0510102993683055e-05,
+      "loss": 0.5635,
+      "r_loss": 0.0,
+      "step": 3690,
+      "steps": 3690
+    },
+    {
+      "epoch": 0.493,
+      "learning_rate": 1.0482651744026366e-05,
+      "loss": 0.5601,
+      "r_loss": 0.0,
+      "step": 3700,
+      "steps": 3700
+    },
+    {
+      "epoch": 0.494,
+      "learning_rate": 1.0455200494369678e-05,
+      "loss": 0.5665,
+      "r_loss": 0.0,
+      "step": 3710,
+      "steps": 3710
+    },
+    {
+      "epoch": 0.495,
+      "learning_rate": 1.0427749244712991e-05,
+      "loss": 0.5684,
+      "r_loss": 0.0,
+      "step": 3720,
+      "steps": 3720
+    },
+    {
+      "epoch": 0.497,
+      "learning_rate": 1.0400297995056303e-05,
+      "loss": 0.577,
+      "r_loss": 0.0,
+      "step": 3730,
+      "steps": 3730
+    },
+    {
+      "epoch": 0.498,
+      "learning_rate": 1.0372846745399615e-05,
+      "loss": 0.563,
+      "r_loss": 0.0,
+      "step": 3740,
+      "steps": 3740
+    },
+    {
+      "epoch": 0.499,
+      "learning_rate": 1.0345395495742928e-05,
+      "loss": 0.5563,
+      "r_loss": 0.0,
+      "step": 3750,
+      "steps": 3750
+    },
+    {
+      "epoch": 0.501,
+      "learning_rate": 1.031794424608624e-05,
+      "loss": 0.57,
+      "r_loss": 0.0,
+      "step": 3760,
+      "steps": 3760
+    },
+    {
+      "epoch": 0.502,
+      "learning_rate": 1.0290492996429551e-05,
+      "loss": 0.5668,
+      "r_loss": 0.0,
+      "step": 3770,
+      "steps": 3770
+    },
+    {
+      "epoch": 0.503,
+      "learning_rate": 1.0263041746772864e-05,
+      "loss": 0.5699,
+      "r_loss": 0.0,
+      "step": 3780,
+      "steps": 3780
+    },
+    {
+      "epoch": 0.505,
+      "learning_rate": 1.0235590497116176e-05,
+      "loss": 0.5575,
+      "r_loss": 0.0,
+      "step": 3790,
+      "steps": 3790
+    },
+    {
+      "epoch": 0.506,
+      "learning_rate": 1.0208139247459488e-05,
+      "loss": 0.5555,
+      "r_loss": 0.0,
+      "step": 3800,
+      "steps": 3800
+    },
+    {
+      "epoch": 0.507,
+      "learning_rate": 1.0180687997802801e-05,
+      "loss": 0.5718,
+      "r_loss": 0.0,
+      "step": 3810,
+      "steps": 3810
+    },
+    {
+      "epoch": 0.509,
+      "learning_rate": 1.0153236748146113e-05,
+      "loss": 0.5672,
+      "r_loss": 0.0,
+      "step": 3820,
+      "steps": 3820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0125785498489424e-05,
+      "loss": 0.5606,
+      "r_loss": 0.0,
+      "step": 3830,
+      "steps": 3830
+    },
+    {
+      "epoch": 0.511,
+      "learning_rate": 1.009833424883274e-05,
+      "loss": 0.5656,
+      "r_loss": 0.0,
+      "step": 3840,
+      "steps": 3840
+    },
+    {
+      "epoch": 0.513,
+      "learning_rate": 1.007088299917605e-05,
+      "loss": 0.5578,
+      "r_loss": 0.0,
+      "step": 3850,
+      "steps": 3850
+    },
+    {
+      "epoch": 0.514,
+      "learning_rate": 1.0043431749519364e-05,
+      "loss": 0.5626,
+      "r_loss": 0.0,
+      "step": 3860,
+      "steps": 3860
+    },
+    {
+      "epoch": 0.515,
+      "learning_rate": 1.0015980499862676e-05,
+      "loss": 0.5612,
+      "r_loss": 0.0,
+      "step": 3870,
+      "steps": 3870
+    },
+    {
+      "epoch": 0.517,
+      "learning_rate": 9.988529250205988e-06,
+      "loss": 0.5534,
+      "r_loss": 0.0,
+      "step": 3880,
+      "steps": 3880
+    },
+    {
+      "epoch": 0.518,
+      "learning_rate": 9.9610780005493e-06,
+      "loss": 0.5467,
+      "r_loss": 0.0,
+      "step": 3890,
+      "steps": 3890
+    },
+    {
+      "epoch": 0.519,
+      "learning_rate": 9.933626750892613e-06,
+      "loss": 0.5573,
+      "r_loss": 0.0,
+      "step": 3900,
+      "steps": 3900
+    },
+    {
+      "epoch": 0.521,
+      "learning_rate": 9.906175501235924e-06,
+      "loss": 0.5642,
+      "r_loss": 0.0,
+      "step": 3910,
+      "steps": 3910
+    },
+    {
+      "epoch": 0.522,
+      "learning_rate": 9.878724251579236e-06,
+      "loss": 0.5621,
+      "r_loss": 0.0,
+      "step": 3920,
+      "steps": 3920
+    },
+    {
+      "epoch": 0.523,
+      "learning_rate": 9.851273001922549e-06,
+      "loss": 0.5609,
+      "r_loss": 0.0,
+      "step": 3930,
+      "steps": 3930
+    },
+    {
+      "epoch": 0.525,
+      "learning_rate": 9.82382175226586e-06,
+      "loss": 0.5832,
+      "r_loss": 0.0,
+      "step": 3940,
+      "steps": 3940
+    },
+    {
+      "epoch": 0.526,
+      "learning_rate": 9.796370502609172e-06,
+      "loss": 0.5658,
+      "r_loss": 0.0,
+      "step": 3950,
+      "steps": 3950
+    },
+    {
+      "epoch": 0.527,
+      "learning_rate": 9.768919252952486e-06,
+      "loss": 0.56,
+      "r_loss": 0.0,
+      "step": 3960,
+      "steps": 3960
+    },
+    {
+      "epoch": 0.529,
+      "learning_rate": 9.741468003295797e-06,
+      "loss": 0.5683,
+      "r_loss": 0.0,
+      "step": 3970,
+      "steps": 3970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.714016753639109e-06,
+      "loss": 0.5619,
+      "r_loss": 0.0,
+      "step": 3980,
+      "steps": 3980
+    },
+    {
+      "epoch": 0.531,
+      "learning_rate": 9.686565503982422e-06,
+      "loss": 0.5648,
+      "r_loss": 0.0,
+      "step": 3990,
+      "steps": 3990
+    },
+    {
+      "epoch": 0.533,
+      "learning_rate": 9.659114254325734e-06,
+      "loss": 0.5659,
+      "r_loss": 0.0,
+      "step": 4000,
+      "steps": 4000
+    },
+    {
+      "epoch": 0.534,
+      "learning_rate": 9.631663004669047e-06,
+      "loss": 0.5652,
+      "r_loss": 0.0,
+      "step": 4010,
+      "steps": 4010
+    },
+    {
+      "epoch": 0.535,
+      "learning_rate": 9.604211755012359e-06,
+      "loss": 0.5475,
+      "r_loss": 0.0,
+      "step": 4020,
+      "steps": 4020
+    },
+    {
+      "epoch": 0.537,
+      "learning_rate": 9.576760505355672e-06,
+      "loss": 0.5733,
+      "r_loss": 0.0,
+      "step": 4030,
+      "steps": 4030
+    },
+    {
+      "epoch": 0.538,
+      "learning_rate": 9.549309255698984e-06,
+      "loss": 0.5537,
+      "r_loss": 0.0,
+      "step": 4040,
+      "steps": 4040
+    },
+    {
+      "epoch": 0.539,
+      "learning_rate": 9.521858006042296e-06,
+      "loss": 0.5582,
+      "r_loss": 0.0,
+      "step": 4050,
+      "steps": 4050
+    },
+    {
+      "epoch": 0.541,
+      "learning_rate": 9.494406756385609e-06,
+      "loss": 0.5559,
+      "r_loss": 0.0,
+      "step": 4060,
+      "steps": 4060
+    },
+    {
+      "epoch": 0.542,
+      "learning_rate": 9.46695550672892e-06,
+      "loss": 0.5698,
+      "r_loss": 0.0,
+      "step": 4070,
+      "steps": 4070
+    },
+    {
+      "epoch": 0.543,
+      "learning_rate": 9.439504257072232e-06,
+      "loss": 0.5478,
+      "r_loss": 0.0,
+      "step": 4080,
+      "steps": 4080
+    },
+    {
+      "epoch": 0.545,
+      "learning_rate": 9.412053007415545e-06,
+      "loss": 0.5657,
+      "r_loss": 0.0,
+      "step": 4090,
+      "steps": 4090
+    },
+    {
+      "epoch": 0.546,
+      "learning_rate": 9.384601757758857e-06,
+      "loss": 0.5695,
+      "r_loss": 0.0,
+      "step": 4100,
+      "steps": 4100
+    },
+    {
+      "epoch": 0.547,
+      "learning_rate": 9.35715050810217e-06,
+      "loss": 0.5614,
+      "r_loss": 0.0,
+      "step": 4110,
+      "steps": 4110
+    },
+    {
+      "epoch": 0.549,
+      "learning_rate": 9.329699258445482e-06,
+      "loss": 0.5641,
+      "r_loss": 0.0,
+      "step": 4120,
+      "steps": 4120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.302248008788794e-06,
+      "loss": 0.5645,
+      "r_loss": 0.0,
+      "step": 4130,
+      "steps": 4130
+    },
+    {
+      "epoch": 0.551,
+      "learning_rate": 9.274796759132107e-06,
+      "loss": 0.5613,
+      "r_loss": 0.0,
+      "step": 4140,
+      "steps": 4140
+    },
+    {
+      "epoch": 0.553,
+      "learning_rate": 9.247345509475419e-06,
+      "loss": 0.5482,
+      "r_loss": 0.0,
+      "step": 4150,
+      "steps": 4150
+    },
+    {
+      "epoch": 0.554,
+      "learning_rate": 9.21989425981873e-06,
+      "loss": 0.5579,
+      "r_loss": 0.0,
+      "step": 4160,
+      "steps": 4160
+    },
+    {
+      "epoch": 0.555,
+      "learning_rate": 9.192443010162044e-06,
+      "loss": 0.571,
+      "r_loss": 0.0,
+      "step": 4170,
+      "steps": 4170
+    },
+    {
+      "epoch": 0.557,
+      "learning_rate": 9.164991760505357e-06,
+      "loss": 0.5565,
+      "r_loss": 0.0,
+      "step": 4180,
+      "steps": 4180
+    },
+    {
+      "epoch": 0.558,
+      "learning_rate": 9.137540510848667e-06,
+      "loss": 0.5586,
+      "r_loss": 0.0,
+      "step": 4190,
+      "steps": 4190
+    },
+    {
+      "epoch": 0.559,
+      "learning_rate": 9.11008926119198e-06,
+      "loss": 0.554,
+      "r_loss": 0.0,
+      "step": 4200,
+      "steps": 4200
+    },
+    {
+      "epoch": 0.561,
+      "learning_rate": 9.082638011535293e-06,
+      "loss": 0.5611,
+      "r_loss": 0.0,
+      "step": 4210,
+      "steps": 4210
+    },
+    {
+      "epoch": 0.562,
+      "learning_rate": 9.055186761878605e-06,
+      "loss": 0.5577,
+      "r_loss": 0.0,
+      "step": 4220,
+      "steps": 4220
+    },
+    {
+      "epoch": 0.563,
+      "learning_rate": 9.027735512221917e-06,
+      "loss": 0.563,
+      "r_loss": 0.0,
+      "step": 4230,
+      "steps": 4230
+    },
+    {
+      "epoch": 0.565,
+      "learning_rate": 9.00028426256523e-06,
+      "loss": 0.5642,
+      "r_loss": 0.0,
+      "step": 4240,
+      "steps": 4240
+    },
+    {
+      "epoch": 0.566,
+      "learning_rate": 8.972833012908542e-06,
+      "loss": 0.5513,
+      "r_loss": 0.0,
+      "step": 4250,
+      "steps": 4250
+    },
+    {
+      "epoch": 0.567,
+      "learning_rate": 8.945381763251853e-06,
+      "loss": 0.5591,
+      "r_loss": 0.0,
+      "step": 4260,
+      "steps": 4260
+    },
+    {
+      "epoch": 0.569,
+      "learning_rate": 8.917930513595167e-06,
+      "loss": 0.5575,
+      "r_loss": 0.0,
+      "step": 4270,
+      "steps": 4270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.890479263938478e-06,
+      "loss": 0.5618,
+      "r_loss": 0.0,
+      "step": 4280,
+      "steps": 4280
+    },
+    {
+      "epoch": 0.571,
+      "learning_rate": 8.86302801428179e-06,
+      "loss": 0.5623,
+      "r_loss": 0.0,
+      "step": 4290,
+      "steps": 4290
+    },
+    {
+      "epoch": 0.573,
+      "learning_rate": 8.835576764625103e-06,
+      "loss": 0.5611,
+      "r_loss": 0.0,
+      "step": 4300,
+      "steps": 4300
+    },
+    {
+      "epoch": 0.574,
+      "learning_rate": 8.808125514968415e-06,
+      "loss": 0.5552,
+      "r_loss": 0.0,
+      "step": 4310,
+      "steps": 4310
+    },
+    {
+      "epoch": 0.575,
+      "learning_rate": 8.780674265311727e-06,
+      "loss": 0.556,
+      "r_loss": 0.0,
+      "step": 4320,
+      "steps": 4320
+    },
+    {
+      "epoch": 0.577,
+      "learning_rate": 8.75322301565504e-06,
+      "loss": 0.5585,
+      "r_loss": 0.0,
+      "step": 4330,
+      "steps": 4330
+    },
+    {
+      "epoch": 0.578,
+      "learning_rate": 8.725771765998351e-06,
+      "loss": 0.5637,
+      "r_loss": 0.0,
+      "step": 4340,
+      "steps": 4340
+    },
+    {
+      "epoch": 0.579,
+      "learning_rate": 8.698320516341663e-06,
+      "loss": 0.5596,
+      "r_loss": 0.0,
+      "step": 4350,
+      "steps": 4350
+    },
+    {
+      "epoch": 0.581,
+      "learning_rate": 8.670869266684976e-06,
+      "loss": 0.5531,
+      "r_loss": 0.0,
+      "step": 4360,
+      "steps": 4360
+    },
+    {
+      "epoch": 0.582,
+      "learning_rate": 8.64341801702829e-06,
+      "loss": 0.5545,
+      "r_loss": 0.0,
+      "step": 4370,
+      "steps": 4370
+    },
+    {
+      "epoch": 0.583,
+      "learning_rate": 8.615966767371601e-06,
+      "loss": 0.5587,
+      "r_loss": 0.0,
+      "step": 4380,
+      "steps": 4380
+    },
+    {
+      "epoch": 0.585,
+      "learning_rate": 8.588515517714913e-06,
+      "loss": 0.5584,
+      "r_loss": 0.0,
+      "step": 4390,
+      "steps": 4390
+    },
+    {
+      "epoch": 0.586,
+      "learning_rate": 8.561064268058226e-06,
+      "loss": 0.5572,
+      "r_loss": 0.0,
+      "step": 4400,
+      "steps": 4400
+    },
+    {
+      "epoch": 0.587,
+      "learning_rate": 8.533613018401538e-06,
+      "loss": 0.5581,
+      "r_loss": 0.0,
+      "step": 4410,
+      "steps": 4410
+    },
+    {
+      "epoch": 0.589,
+      "learning_rate": 8.50616176874485e-06,
+      "loss": 0.5568,
+      "r_loss": 0.0,
+      "step": 4420,
+      "steps": 4420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.478710519088163e-06,
+      "loss": 0.5571,
+      "r_loss": 0.0,
+      "step": 4430,
+      "steps": 4430
+    },
+    {
+      "epoch": 0.591,
+      "learning_rate": 8.451259269431475e-06,
+      "loss": 0.5591,
+      "r_loss": 0.0,
+      "step": 4440,
+      "steps": 4440
+    },
+    {
+      "epoch": 0.593,
+      "learning_rate": 8.423808019774786e-06,
+      "loss": 0.5545,
+      "r_loss": 0.0,
+      "step": 4450,
+      "steps": 4450
+    },
+    {
+      "epoch": 0.594,
+      "learning_rate": 8.3963567701181e-06,
+      "loss": 0.5566,
+      "r_loss": 0.0,
+      "step": 4460,
+      "steps": 4460
+    },
+    {
+      "epoch": 0.595,
+      "learning_rate": 8.368905520461411e-06,
+      "loss": 0.5478,
+      "r_loss": 0.0,
+      "step": 4470,
+      "steps": 4470
+    },
+    {
+      "epoch": 0.597,
+      "learning_rate": 8.341454270804725e-06,
+      "loss": 0.5597,
+      "r_loss": 0.0,
+      "step": 4480,
+      "steps": 4480
+    },
+    {
+      "epoch": 0.598,
+      "learning_rate": 8.314003021148036e-06,
+      "loss": 0.5471,
+      "r_loss": 0.0,
+      "step": 4490,
+      "steps": 4490
+    },
+    {
+      "epoch": 0.599,
+      "learning_rate": 8.286551771491348e-06,
+      "loss": 0.5528,
+      "r_loss": 0.0,
+      "step": 4500,
+      "steps": 4500
+    },
+    {
+      "epoch": 0.601,
+      "learning_rate": 8.259100521834661e-06,
+      "loss": 0.5452,
+      "r_loss": 0.0,
+      "step": 4510,
+      "steps": 4510
+    },
+    {
+      "epoch": 0.602,
+      "learning_rate": 8.231649272177973e-06,
+      "loss": 0.5621,
+      "r_loss": 0.0,
+      "step": 4520,
+      "steps": 4520
+    },
+    {
+      "epoch": 0.603,
+      "learning_rate": 8.204198022521284e-06,
+      "loss": 0.5506,
+      "r_loss": 0.0,
+      "step": 4530,
+      "steps": 4530
+    },
+    {
+      "epoch": 0.605,
+      "learning_rate": 8.176746772864598e-06,
+      "loss": 0.556,
+      "r_loss": 0.0,
+      "step": 4540,
+      "steps": 4540
+    },
+    {
+      "epoch": 0.606,
+      "learning_rate": 8.149295523207911e-06,
+      "loss": 0.5507,
+      "r_loss": 0.0,
+      "step": 4550,
+      "steps": 4550
+    },
+    {
+      "epoch": 0.607,
+      "learning_rate": 8.121844273551223e-06,
+      "loss": 0.5598,
+      "r_loss": 0.0,
+      "step": 4560,
+      "steps": 4560
+    },
+    {
+      "epoch": 0.609,
+      "learning_rate": 8.094393023894534e-06,
+      "loss": 0.5581,
+      "r_loss": 0.0,
+      "step": 4570,
+      "steps": 4570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.066941774237848e-06,
+      "loss": 0.5487,
+      "r_loss": 0.0,
+      "step": 4580,
+      "steps": 4580
+    },
+    {
+      "epoch": 0.611,
+      "learning_rate": 8.03949052458116e-06,
+      "loss": 0.5549,
+      "r_loss": 0.0,
+      "step": 4590,
+      "steps": 4590
+    },
+    {
+      "epoch": 0.613,
+      "learning_rate": 8.012039274924471e-06,
+      "loss": 0.5531,
+      "r_loss": 0.0,
+      "step": 4600,
+      "steps": 4600
+    },
+    {
+      "epoch": 0.614,
+      "learning_rate": 7.984588025267784e-06,
+      "loss": 0.5548,
+      "r_loss": 0.0,
+      "step": 4610,
+      "steps": 4610
+    },
+    {
+      "epoch": 0.615,
+      "learning_rate": 7.957136775611096e-06,
+      "loss": 0.5549,
+      "r_loss": 0.0,
+      "step": 4620,
+      "steps": 4620
+    },
+    {
+      "epoch": 0.617,
+      "learning_rate": 7.935175775885745e-06,
+      "loss": 0.5634,
+      "r_loss": 0.0,
+      "step": 4630,
+      "steps": 4630
+    },
+    {
+      "epoch": 0.618,
+      "learning_rate": 7.907724526229059e-06,
+      "loss": 0.5514,
+      "r_loss": 0.0,
+      "step": 4640,
+      "steps": 4640
+    },
+    {
+      "epoch": 0.619,
+      "learning_rate": 7.88027327657237e-06,
+      "loss": 0.56,
+      "r_loss": 0.0,
+      "step": 4650,
+      "steps": 4650
+    },
+    {
+      "epoch": 0.621,
+      "learning_rate": 7.852822026915682e-06,
+      "loss": 0.544,
+      "r_loss": 0.0,
+      "step": 4660,
+      "steps": 4660
+    },
+    {
+      "epoch": 0.622,
+      "learning_rate": 7.825370777258995e-06,
+      "loss": 0.5523,
+      "r_loss": 0.0,
+      "step": 4670,
+      "steps": 4670
+    },
+    {
+      "epoch": 0.623,
+      "learning_rate": 7.797919527602307e-06,
+      "loss": 0.5553,
+      "r_loss": 0.0,
+      "step": 4680,
+      "steps": 4680
+    },
+    {
+      "epoch": 0.625,
+      "learning_rate": 7.770468277945619e-06,
+      "loss": 0.5566,
+      "r_loss": 0.0,
+      "step": 4690,
+      "steps": 4690
+    },
+    {
+      "epoch": 0.626,
+      "learning_rate": 7.743017028288932e-06,
+      "loss": 0.5543,
+      "r_loss": 0.0,
+      "step": 4700,
+      "steps": 4700
+    },
+    {
+      "epoch": 0.627,
+      "learning_rate": 7.715565778632244e-06,
+      "loss": 0.555,
+      "r_loss": 0.0,
+      "step": 4710,
+      "steps": 4710
+    },
+    {
+      "epoch": 0.629,
+      "learning_rate": 7.688114528975555e-06,
+      "loss": 0.5407,
+      "r_loss": 0.0,
+      "step": 4720,
+      "steps": 4720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.660663279318869e-06,
+      "loss": 0.5538,
+      "r_loss": 0.0,
+      "step": 4730,
+      "steps": 4730
+    },
+    {
+      "epoch": 0.631,
+      "learning_rate": 7.633212029662182e-06,
+      "loss": 0.559,
+      "r_loss": 0.0,
+      "step": 4740,
+      "steps": 4740
+    },
+    {
+      "epoch": 0.633,
+      "learning_rate": 7.605760780005493e-06,
+      "loss": 0.5467,
+      "r_loss": 0.0,
+      "step": 4750,
+      "steps": 4750
+    },
+    {
+      "epoch": 0.634,
+      "learning_rate": 7.578309530348805e-06,
+      "loss": 0.5422,
+      "r_loss": 0.0,
+      "step": 4760,
+      "steps": 4760
+    },
+    {
+      "epoch": 0.635,
+      "learning_rate": 7.550858280692118e-06,
+      "loss": 0.5469,
+      "r_loss": 0.0,
+      "step": 4770,
+      "steps": 4770
+    },
+    {
+      "epoch": 0.637,
+      "learning_rate": 7.52340703103543e-06,
+      "loss": 0.5371,
+      "r_loss": 0.0,
+      "step": 4780,
+      "steps": 4780
+    },
+    {
+      "epoch": 0.638,
+      "learning_rate": 7.495955781378742e-06,
+      "loss": 0.565,
+      "r_loss": 0.0,
+      "step": 4790,
+      "steps": 4790
+    },
+    {
+      "epoch": 0.639,
+      "learning_rate": 7.468504531722054e-06,
+      "loss": 0.5323,
+      "r_loss": 0.0,
+      "step": 4800,
+      "steps": 4800
+    },
+    {
+      "epoch": 0.641,
+      "learning_rate": 7.441053282065367e-06,
+      "loss": 0.5507,
+      "r_loss": 0.0,
+      "step": 4810,
+      "steps": 4810
+    },
+    {
+      "epoch": 0.642,
+      "learning_rate": 7.413602032408678e-06,
+      "loss": 0.5496,
+      "r_loss": 0.0,
+      "step": 4820,
+      "steps": 4820
+    },
+    {
+      "epoch": 0.643,
+      "learning_rate": 7.386150782751991e-06,
+      "loss": 0.5447,
+      "r_loss": 0.0,
+      "step": 4830,
+      "steps": 4830
+    },
+    {
+      "epoch": 0.645,
+      "learning_rate": 7.358699533095303e-06,
+      "loss": 0.539,
+      "r_loss": 0.0,
+      "step": 4840,
+      "steps": 4840
+    },
+    {
+      "epoch": 0.646,
+      "learning_rate": 7.331248283438615e-06,
+      "loss": 0.5462,
+      "r_loss": 0.0,
+      "step": 4850,
+      "steps": 4850
+    },
+    {
+      "epoch": 0.647,
+      "learning_rate": 7.3037970337819274e-06,
+      "loss": 0.5478,
+      "r_loss": 0.0,
+      "step": 4860,
+      "steps": 4860
+    },
+    {
+      "epoch": 0.649,
+      "learning_rate": 7.276345784125241e-06,
+      "loss": 0.5394,
+      "r_loss": 0.0,
+      "step": 4870,
+      "steps": 4870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.248894534468553e-06,
+      "loss": 0.5507,
+      "r_loss": 0.0,
+      "step": 4880,
+      "steps": 4880
+    },
+    {
+      "epoch": 0.651,
+      "learning_rate": 7.221443284811865e-06,
+      "loss": 0.5422,
+      "r_loss": 0.0,
+      "step": 4890,
+      "steps": 4890
+    },
+    {
+      "epoch": 0.653,
+      "learning_rate": 7.193992035155177e-06,
+      "loss": 0.555,
+      "r_loss": 0.0,
+      "step": 4900,
+      "steps": 4900
+    },
+    {
+      "epoch": 0.654,
+      "learning_rate": 7.16654078549849e-06,
+      "loss": 0.5385,
+      "r_loss": 0.0,
+      "step": 4910,
+      "steps": 4910
+    },
+    {
+      "epoch": 0.655,
+      "learning_rate": 7.1390895358418015e-06,
+      "loss": 0.5517,
+      "r_loss": 0.0,
+      "step": 4920,
+      "steps": 4920
+    },
+    {
+      "epoch": 0.657,
+      "learning_rate": 7.111638286185114e-06,
+      "loss": 0.5419,
+      "r_loss": 0.0,
+      "step": 4930,
+      "steps": 4930
+    },
+    {
+      "epoch": 0.658,
+      "learning_rate": 7.0841870365284264e-06,
+      "loss": 0.5396,
+      "r_loss": 0.0,
+      "step": 4940,
+      "steps": 4940
+    },
+    {
+      "epoch": 0.659,
+      "learning_rate": 7.056735786871739e-06,
+      "loss": 0.5458,
+      "r_loss": 0.0,
+      "step": 4950,
+      "steps": 4950
+    },
+    {
+      "epoch": 0.661,
+      "learning_rate": 7.0292845372150506e-06,
+      "loss": 0.5462,
+      "r_loss": 0.0,
+      "step": 4960,
+      "steps": 4960
+    },
+    {
+      "epoch": 0.662,
+      "learning_rate": 7.001833287558363e-06,
+      "loss": 0.5538,
+      "r_loss": 0.0,
+      "step": 4970,
+      "steps": 4970
+    },
+    {
+      "epoch": 0.663,
+      "learning_rate": 6.9743820379016755e-06,
+      "loss": 0.5599,
+      "r_loss": 0.0,
+      "step": 4980,
+      "steps": 4980
+    },
+    {
+      "epoch": 0.665,
+      "learning_rate": 6.946930788244987e-06,
+      "loss": 0.5595,
+      "r_loss": 0.0,
+      "step": 4990,
+      "steps": 4990
+    },
+    {
+      "epoch": 0.666,
+      "learning_rate": 6.9194795385883e-06,
+      "loss": 0.5373,
+      "r_loss": 0.0,
+      "step": 5000,
+      "steps": 5000
+    },
+    {
+      "epoch": 0.667,
+      "learning_rate": 6.892028288931612e-06,
+      "loss": 0.5513,
+      "r_loss": 0.0,
+      "step": 5010,
+      "steps": 5010
+    },
+    {
+      "epoch": 0.669,
+      "learning_rate": 6.864577039274924e-06,
+      "loss": 0.5394,
+      "r_loss": 0.0,
+      "step": 5020,
+      "steps": 5020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.837125789618236e-06,
+      "loss": 0.5605,
+      "r_loss": 0.0,
+      "step": 5030,
+      "steps": 5030
+    },
+    {
+      "epoch": 0.671,
+      "learning_rate": 6.809674539961549e-06,
+      "loss": 0.5457,
+      "r_loss": 0.0,
+      "step": 5040,
+      "steps": 5040
+    },
+    {
+      "epoch": 0.673,
+      "learning_rate": 6.782223290304862e-06,
+      "loss": 0.5463,
+      "r_loss": 0.0,
+      "step": 5050,
+      "steps": 5050
+    },
+    {
+      "epoch": 0.674,
+      "learning_rate": 6.754772040648173e-06,
+      "loss": 0.5446,
+      "r_loss": 0.0,
+      "step": 5060,
+      "steps": 5060
+    },
+    {
+      "epoch": 0.675,
+      "learning_rate": 6.727320790991486e-06,
+      "loss": 0.5386,
+      "r_loss": 0.0,
+      "step": 5070,
+      "steps": 5070
+    },
+    {
+      "epoch": 0.677,
+      "learning_rate": 6.699869541334799e-06,
+      "loss": 0.5416,
+      "r_loss": 0.0,
+      "step": 5080,
+      "steps": 5080
+    },
+    {
+      "epoch": 0.678,
+      "learning_rate": 6.67241829167811e-06,
+      "loss": 0.5375,
+      "r_loss": 0.0,
+      "step": 5090,
+      "steps": 5090
+    },
+    {
+      "epoch": 0.679,
+      "learning_rate": 6.644967042021423e-06,
+      "loss": 0.5425,
+      "r_loss": 0.0,
+      "step": 5100,
+      "steps": 5100
+    },
+    {
+      "epoch": 0.681,
+      "learning_rate": 6.617515792364735e-06,
+      "loss": 0.5397,
+      "r_loss": 0.0,
+      "step": 5110,
+      "steps": 5110
+    },
+    {
+      "epoch": 0.682,
+      "learning_rate": 6.590064542708047e-06,
+      "loss": 0.5337,
+      "r_loss": 0.0,
+      "step": 5120,
+      "steps": 5120
+    },
+    {
+      "epoch": 0.683,
+      "learning_rate": 6.562613293051359e-06,
+      "loss": 0.5402,
+      "r_loss": 0.0,
+      "step": 5130,
+      "steps": 5130
+    },
+    {
+      "epoch": 0.685,
+      "learning_rate": 6.535162043394672e-06,
+      "loss": 0.5358,
+      "r_loss": 0.0,
+      "step": 5140,
+      "steps": 5140
+    },
+    {
+      "epoch": 0.686,
+      "learning_rate": 6.507710793737984e-06,
+      "loss": 0.5458,
+      "r_loss": 0.0,
+      "step": 5150,
+      "steps": 5150
+    },
+    {
+      "epoch": 0.687,
+      "learning_rate": 6.480259544081296e-06,
+      "loss": 0.5467,
+      "r_loss": 0.0,
+      "step": 5160,
+      "steps": 5160
+    },
+    {
+      "epoch": 0.689,
+      "learning_rate": 6.452808294424608e-06,
+      "loss": 0.5457,
+      "r_loss": 0.0,
+      "step": 5170,
+      "steps": 5170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.425357044767921e-06,
+      "loss": 0.5511,
+      "r_loss": 0.0,
+      "step": 5180,
+      "steps": 5180
+    },
+    {
+      "epoch": 0.691,
+      "learning_rate": 6.3979057951112325e-06,
+      "loss": 0.5491,
+      "r_loss": 0.0,
+      "step": 5190,
+      "steps": 5190
+    },
+    {
+      "epoch": 0.693,
+      "learning_rate": 6.370454545454545e-06,
+      "loss": 0.5306,
+      "r_loss": 0.0,
+      "step": 5200,
+      "steps": 5200
+    },
+    {
+      "epoch": 0.694,
+      "learning_rate": 6.3430032957978575e-06,
+      "loss": 0.5473,
+      "r_loss": 0.0,
+      "step": 5210,
+      "steps": 5210
+    },
+    {
+      "epoch": 0.695,
+      "learning_rate": 6.315552046141169e-06,
+      "loss": 0.5433,
+      "r_loss": 0.0,
+      "step": 5220,
+      "steps": 5220
+    },
+    {
+      "epoch": 0.697,
+      "learning_rate": 6.288100796484482e-06,
+      "loss": 0.5413,
+      "r_loss": 0.0,
+      "step": 5230,
+      "steps": 5230
+    },
+    {
+      "epoch": 0.698,
+      "learning_rate": 6.260649546827795e-06,
+      "loss": 0.5395,
+      "r_loss": 0.0,
+      "step": 5240,
+      "steps": 5240
+    },
+    {
+      "epoch": 0.699,
+      "learning_rate": 6.233198297171107e-06,
+      "loss": 0.54,
+      "r_loss": 0.0,
+      "step": 5250,
+      "steps": 5250
+    },
+    {
+      "epoch": 0.701,
+      "learning_rate": 6.205747047514419e-06,
+      "loss": 0.5302,
+      "r_loss": 0.0,
+      "step": 5260,
+      "steps": 5260
+    },
+    {
+      "epoch": 0.702,
+      "learning_rate": 6.1782957978577315e-06,
+      "loss": 0.5379,
+      "r_loss": 0.0,
+      "step": 5270,
+      "steps": 5270
+    },
+    {
+      "epoch": 0.703,
+      "learning_rate": 6.150844548201044e-06,
+      "loss": 0.5371,
+      "r_loss": 0.0,
+      "step": 5280,
+      "steps": 5280
+    },
+    {
+      "epoch": 0.705,
+      "learning_rate": 6.123393298544356e-06,
+      "loss": 0.5464,
+      "r_loss": 0.0,
+      "step": 5290,
+      "steps": 5290
+    },
+    {
+      "epoch": 0.706,
+      "learning_rate": 6.095942048887668e-06,
+      "loss": 0.5412,
+      "r_loss": 0.0,
+      "step": 5300,
+      "steps": 5300
+    },
+    {
+      "epoch": 0.707,
+      "learning_rate": 6.0684907992309806e-06,
+      "loss": 0.5414,
+      "r_loss": 0.0,
+      "step": 5310,
+      "steps": 5310
+    },
+    {
+      "epoch": 0.709,
+      "learning_rate": 6.041039549574293e-06,
+      "loss": 0.5508,
+      "r_loss": 0.0,
+      "step": 5320,
+      "steps": 5320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.013588299917605e-06,
+      "loss": 0.5484,
+      "r_loss": 0.0,
+      "step": 5330,
+      "steps": 5330
+    },
+    {
+      "epoch": 0.711,
+      "learning_rate": 5.986137050260917e-06,
+      "loss": 0.548,
+      "r_loss": 0.0,
+      "step": 5340,
+      "steps": 5340
+    },
+    {
+      "epoch": 0.713,
+      "learning_rate": 5.95868580060423e-06,
+      "loss": 0.5386,
+      "r_loss": 0.0,
+      "step": 5350,
+      "steps": 5350
+    },
+    {
+      "epoch": 0.714,
+      "learning_rate": 5.931234550947541e-06,
+      "loss": 0.534,
+      "r_loss": 0.0,
+      "step": 5360,
+      "steps": 5360
+    },
+    {
+      "epoch": 0.715,
+      "learning_rate": 5.903783301290854e-06,
+      "loss": 0.5256,
+      "r_loss": 0.0,
+      "step": 5370,
+      "steps": 5370
+    },
+    {
+      "epoch": 0.717,
+      "learning_rate": 5.876332051634166e-06,
+      "loss": 0.5396,
+      "r_loss": 0.0,
+      "step": 5380,
+      "steps": 5380
+    },
+    {
+      "epoch": 0.718,
+      "learning_rate": 5.848880801977478e-06,
+      "loss": 0.5547,
+      "r_loss": 0.0,
+      "step": 5390,
+      "steps": 5390
+    },
+    {
+      "epoch": 0.719,
+      "learning_rate": 5.82142955232079e-06,
+      "loss": 0.5372,
+      "r_loss": 0.0,
+      "step": 5400,
+      "steps": 5400
+    },
+    {
+      "epoch": 0.721,
+      "learning_rate": 5.793978302664104e-06,
+      "loss": 0.5345,
+      "r_loss": 0.0,
+      "step": 5410,
+      "steps": 5410
+    },
+    {
+      "epoch": 0.722,
+      "learning_rate": 5.766527053007416e-06,
+      "loss": 0.5487,
+      "r_loss": 0.0,
+      "step": 5420,
+      "steps": 5420
+    },
+    {
+      "epoch": 0.723,
+      "learning_rate": 5.739075803350728e-06,
+      "loss": 0.5357,
+      "r_loss": 0.0,
+      "step": 5430,
+      "steps": 5430
+    },
+    {
+      "epoch": 0.725,
+      "learning_rate": 5.71162455369404e-06,
+      "loss": 0.5388,
+      "r_loss": 0.0,
+      "step": 5440,
+      "steps": 5440
+    },
+    {
+      "epoch": 0.726,
+      "learning_rate": 5.684173304037353e-06,
+      "loss": 0.5274,
+      "r_loss": 0.0,
+      "step": 5450,
+      "steps": 5450
+    },
+    {
+      "epoch": 0.727,
+      "learning_rate": 5.656722054380664e-06,
+      "loss": 0.5389,
+      "r_loss": 0.0,
+      "step": 5460,
+      "steps": 5460
+    },
+    {
+      "epoch": 0.729,
+      "learning_rate": 5.629270804723977e-06,
+      "loss": 0.5447,
+      "r_loss": 0.0,
+      "step": 5470,
+      "steps": 5470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.601819555067289e-06,
+      "loss": 0.5535,
+      "r_loss": 0.0,
+      "step": 5480,
+      "steps": 5480
+    },
+    {
+      "epoch": 0.731,
+      "learning_rate": 5.574368305410601e-06,
+      "loss": 0.5339,
+      "r_loss": 0.0,
+      "step": 5490,
+      "steps": 5490
+    },
+    {
+      "epoch": 0.733,
+      "learning_rate": 5.5469170557539135e-06,
+      "loss": 0.545,
+      "r_loss": 0.0,
+      "step": 5500,
+      "steps": 5500
+    },
+    {
+      "epoch": 0.734,
+      "learning_rate": 5.519465806097226e-06,
+      "loss": 0.546,
+      "r_loss": 0.0,
+      "step": 5510,
+      "steps": 5510
+    },
+    {
+      "epoch": 0.735,
+      "learning_rate": 5.492014556440538e-06,
+      "loss": 0.5385,
+      "r_loss": 0.0,
+      "step": 5520,
+      "steps": 5520
+    },
+    {
+      "epoch": 0.737,
+      "learning_rate": 5.46456330678385e-06,
+      "loss": 0.5448,
+      "r_loss": 0.0,
+      "step": 5530,
+      "steps": 5530
+    },
+    {
+      "epoch": 0.738,
+      "learning_rate": 5.4371120571271625e-06,
+      "loss": 0.5352,
+      "r_loss": 0.0,
+      "step": 5540,
+      "steps": 5540
+    },
+    {
+      "epoch": 0.739,
+      "learning_rate": 5.409660807470475e-06,
+      "loss": 0.5359,
+      "r_loss": 0.0,
+      "step": 5550,
+      "steps": 5550
+    },
+    {
+      "epoch": 0.741,
+      "learning_rate": 5.382209557813787e-06,
+      "loss": 0.5318,
+      "r_loss": 0.0,
+      "step": 5560,
+      "steps": 5560
+    },
+    {
+      "epoch": 0.742,
+      "learning_rate": 5.354758308157099e-06,
+      "loss": 0.5354,
+      "r_loss": 0.0,
+      "step": 5570,
+      "steps": 5570
+    },
+    {
+      "epoch": 0.743,
+      "learning_rate": 5.3273070585004124e-06,
+      "loss": 0.5344,
+      "r_loss": 0.0,
+      "step": 5580,
+      "steps": 5580
+    },
+    {
+      "epoch": 0.745,
+      "learning_rate": 5.299855808843725e-06,
+      "loss": 0.5399,
+      "r_loss": 0.0,
+      "step": 5590,
+      "steps": 5590
+    },
+    {
+      "epoch": 0.746,
+      "learning_rate": 5.2724045591870366e-06,
+      "loss": 0.5471,
+      "r_loss": 0.0,
+      "step": 5600,
+      "steps": 5600
+    },
+    {
+      "epoch": 0.747,
+      "learning_rate": 5.244953309530349e-06,
+      "loss": 0.5273,
+      "r_loss": 0.0,
+      "step": 5610,
+      "steps": 5610
+    },
+    {
+      "epoch": 0.749,
+      "learning_rate": 5.2175020598736615e-06,
+      "loss": 0.5429,
+      "r_loss": 0.0,
+      "step": 5620,
+      "steps": 5620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.19554106014831e-06,
+      "loss": 0.5176,
+      "r_loss": 0.0,
+      "step": 5630,
+      "steps": 5630
+    },
+    {
+      "epoch": 0.751,
+      "learning_rate": 5.168089810491624e-06,
+      "loss": 0.5268,
+      "r_loss": 0.0,
+      "step": 5640,
+      "steps": 5640
+    },
+    {
+      "epoch": 0.753,
+      "learning_rate": 5.140638560834936e-06,
+      "loss": 0.5311,
+      "r_loss": 0.0,
+      "step": 5650,
+      "steps": 5650
+    },
+    {
+      "epoch": 0.754,
+      "learning_rate": 5.113187311178247e-06,
+      "loss": 0.5309,
+      "r_loss": 0.0,
+      "step": 5660,
+      "steps": 5660
+    },
+    {
+      "epoch": 0.755,
+      "learning_rate": 5.08573606152156e-06,
+      "loss": 0.5409,
+      "r_loss": 0.0,
+      "step": 5670,
+      "steps": 5670
+    },
+    {
+      "epoch": 0.757,
+      "learning_rate": 5.058284811864873e-06,
+      "loss": 0.5318,
+      "r_loss": 0.0,
+      "step": 5680,
+      "steps": 5680
+    },
+    {
+      "epoch": 0.758,
+      "learning_rate": 5.030833562208184e-06,
+      "loss": 0.545,
+      "r_loss": 0.0,
+      "step": 5690,
+      "steps": 5690
+    },
+    {
+      "epoch": 0.759,
+      "learning_rate": 5.003382312551497e-06,
+      "loss": 0.5413,
+      "r_loss": 0.0,
+      "step": 5700,
+      "steps": 5700
+    },
+    {
+      "epoch": 0.761,
+      "learning_rate": 4.975931062894809e-06,
+      "loss": 0.5307,
+      "r_loss": 0.0,
+      "step": 5710,
+      "steps": 5710
+    },
+    {
+      "epoch": 0.762,
+      "learning_rate": 4.948479813238121e-06,
+      "loss": 0.5491,
+      "r_loss": 0.0,
+      "step": 5720,
+      "steps": 5720
+    },
+    {
+      "epoch": 0.763,
+      "learning_rate": 4.921028563581433e-06,
+      "loss": 0.5411,
+      "r_loss": 0.0,
+      "step": 5730,
+      "steps": 5730
+    },
+    {
+      "epoch": 0.765,
+      "learning_rate": 4.893577313924746e-06,
+      "loss": 0.5456,
+      "r_loss": 0.0,
+      "step": 5740,
+      "steps": 5740
+    },
+    {
+      "epoch": 0.766,
+      "learning_rate": 4.866126064268058e-06,
+      "loss": 0.5505,
+      "r_loss": 0.0,
+      "step": 5750,
+      "steps": 5750
+    },
+    {
+      "epoch": 0.767,
+      "learning_rate": 4.838674814611371e-06,
+      "loss": 0.5441,
+      "r_loss": 0.0,
+      "step": 5760,
+      "steps": 5760
+    },
+    {
+      "epoch": 0.769,
+      "learning_rate": 4.8112235649546825e-06,
+      "loss": 0.5405,
+      "r_loss": 0.0,
+      "step": 5770,
+      "steps": 5770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.786517440263664e-06,
+      "loss": 0.5445,
+      "r_loss": 0.0,
+      "step": 5780,
+      "steps": 5780
+    },
+    {
+      "epoch": 0.771,
+      "learning_rate": 4.7590661906069765e-06,
+      "loss": 0.5338,
+      "r_loss": 0.0,
+      "step": 5790,
+      "steps": 5790
+    },
+    {
+      "epoch": 0.773,
+      "learning_rate": 4.731614940950288e-06,
+      "loss": 0.5334,
+      "r_loss": 0.0,
+      "step": 5800,
+      "steps": 5800
+    },
+    {
+      "epoch": 0.774,
+      "learning_rate": 4.7041636912936006e-06,
+      "loss": 0.5424,
+      "r_loss": 0.0,
+      "step": 5810,
+      "steps": 5810
+    },
+    {
+      "epoch": 0.775,
+      "learning_rate": 4.676712441636913e-06,
+      "loss": 0.5295,
+      "r_loss": 0.0,
+      "step": 5820,
+      "steps": 5820
+    },
+    {
+      "epoch": 0.777,
+      "learning_rate": 4.649261191980225e-06,
+      "loss": 0.5504,
+      "r_loss": 0.0,
+      "step": 5830,
+      "steps": 5830
+    },
+    {
+      "epoch": 0.778,
+      "learning_rate": 4.621809942323537e-06,
+      "loss": 0.5261,
+      "r_loss": 0.0,
+      "step": 5840,
+      "steps": 5840
+    },
+    {
+      "epoch": 0.779,
+      "learning_rate": 4.59435869266685e-06,
+      "loss": 0.5414,
+      "r_loss": 0.0,
+      "step": 5850,
+      "steps": 5850
+    },
+    {
+      "epoch": 0.781,
+      "learning_rate": 4.566907443010162e-06,
+      "loss": 0.5431,
+      "r_loss": 0.0,
+      "step": 5860,
+      "steps": 5860
+    },
+    {
+      "epoch": 0.782,
+      "learning_rate": 4.539456193353475e-06,
+      "loss": 0.5411,
+      "r_loss": 0.0,
+      "step": 5870,
+      "steps": 5870
+    },
+    {
+      "epoch": 0.783,
+      "learning_rate": 4.512004943696786e-06,
+      "loss": 0.5377,
+      "r_loss": 0.0,
+      "step": 5880,
+      "steps": 5880
+    },
+    {
+      "epoch": 0.784,
+      "learning_rate": 4.484553694040099e-06,
+      "loss": 0.525,
+      "r_loss": 0.0,
+      "step": 5890,
+      "steps": 5890
+    },
+    {
+      "epoch": 0.786,
+      "learning_rate": 4.457102444383411e-06,
+      "loss": 0.5329,
+      "r_loss": 0.0,
+      "step": 5900,
+      "steps": 5900
+    },
+    {
+      "epoch": 0.787,
+      "learning_rate": 4.429651194726724e-06,
+      "loss": 0.5438,
+      "r_loss": 0.0,
+      "step": 5910,
+      "steps": 5910
+    },
+    {
+      "epoch": 0.788,
+      "learning_rate": 4.402199945070035e-06,
+      "loss": 0.5307,
+      "r_loss": 0.0,
+      "step": 5920,
+      "steps": 5920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.374748695413348e-06,
+      "loss": 0.5362,
+      "r_loss": 0.0,
+      "step": 5930,
+      "steps": 5930
+    },
+    {
+      "epoch": 0.791,
+      "learning_rate": 4.34729744575666e-06,
+      "loss": 0.5474,
+      "r_loss": 0.0,
+      "step": 5940,
+      "steps": 5940
+    },
+    {
+      "epoch": 0.792,
+      "learning_rate": 4.319846196099972e-06,
+      "loss": 0.5352,
+      "r_loss": 0.0,
+      "step": 5950,
+      "steps": 5950
+    },
+    {
+      "epoch": 0.794,
+      "learning_rate": 4.292394946443285e-06,
+      "loss": 0.5288,
+      "r_loss": 0.0,
+      "step": 5960,
+      "steps": 5960
+    },
+    {
+      "epoch": 0.795,
+      "learning_rate": 4.264943696786597e-06,
+      "loss": 0.5431,
+      "r_loss": 0.0,
+      "step": 5970,
+      "steps": 5970
+    },
+    {
+      "epoch": 0.796,
+      "learning_rate": 4.237492447129909e-06,
+      "loss": 0.5349,
+      "r_loss": 0.0,
+      "step": 5980,
+      "steps": 5980
+    },
+    {
+      "epoch": 0.798,
+      "learning_rate": 4.210041197473222e-06,
+      "loss": 0.5295,
+      "r_loss": 0.0,
+      "step": 5990,
+      "steps": 5990
+    },
+    {
+      "epoch": 0.799,
+      "learning_rate": 4.1825899478165335e-06,
+      "loss": 0.5365,
+      "r_loss": 0.0,
+      "step": 6000,
+      "steps": 6000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.155138698159846e-06,
+      "loss": 0.5356,
+      "r_loss": 0.0,
+      "step": 6010,
+      "steps": 6010
+    },
+    {
+      "epoch": 0.802,
+      "learning_rate": 4.127687448503158e-06,
+      "loss": 0.5356,
+      "r_loss": 0.0,
+      "step": 6020,
+      "steps": 6020
+    },
+    {
+      "epoch": 0.803,
+      "learning_rate": 4.10023619884647e-06,
+      "loss": 0.548,
+      "r_loss": 0.0,
+      "step": 6030,
+      "steps": 6030
+    },
+    {
+      "epoch": 0.804,
+      "learning_rate": 4.072784949189783e-06,
+      "loss": 0.5384,
+      "r_loss": 0.0,
+      "step": 6040,
+      "steps": 6040
+    },
+    {
+      "epoch": 0.806,
+      "learning_rate": 4.045333699533095e-06,
+      "loss": 0.5337,
+      "r_loss": 0.0,
+      "step": 6050,
+      "steps": 6050
+    },
+    {
+      "epoch": 0.807,
+      "learning_rate": 4.0178824498764075e-06,
+      "loss": 0.5406,
+      "r_loss": 0.0,
+      "step": 6060,
+      "steps": 6060
+    },
+    {
+      "epoch": 0.808,
+      "learning_rate": 3.99043120021972e-06,
+      "loss": 0.5345,
+      "r_loss": 0.0,
+      "step": 6070,
+      "steps": 6070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9629799505630324e-06,
+      "loss": 0.5253,
+      "r_loss": 0.0,
+      "step": 6080,
+      "steps": 6080
+    },
+    {
+      "epoch": 0.811,
+      "learning_rate": 3.935528700906344e-06,
+      "loss": 0.531,
+      "r_loss": 0.0,
+      "step": 6090,
+      "steps": 6090
+    },
+    {
+      "epoch": 0.812,
+      "learning_rate": 3.9080774512496566e-06,
+      "loss": 0.5278,
+      "r_loss": 0.0,
+      "step": 6100,
+      "steps": 6100
+    },
+    {
+      "epoch": 0.814,
+      "learning_rate": 3.880626201592969e-06,
+      "loss": 0.5297,
+      "r_loss": 0.0,
+      "step": 6110,
+      "steps": 6110
+    },
+    {
+      "epoch": 0.815,
+      "learning_rate": 3.853174951936281e-06,
+      "loss": 0.5255,
+      "r_loss": 0.0,
+      "step": 6120,
+      "steps": 6120
+    },
+    {
+      "epoch": 0.816,
+      "learning_rate": 3.825723702279594e-06,
+      "loss": 0.5465,
+      "r_loss": 0.0,
+      "step": 6130,
+      "steps": 6130
+    },
+    {
+      "epoch": 0.818,
+      "learning_rate": 3.7982724526229056e-06,
+      "loss": 0.5374,
+      "r_loss": 0.0,
+      "step": 6140,
+      "steps": 6140
+    },
+    {
+      "epoch": 0.819,
+      "learning_rate": 3.7708212029662177e-06,
+      "loss": 0.5264,
+      "r_loss": 0.0,
+      "step": 6150,
+      "steps": 6150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7433699533095306e-06,
+      "loss": 0.5273,
+      "r_loss": 0.0,
+      "step": 6160,
+      "steps": 6160
+    },
+    {
+      "epoch": 0.822,
+      "learning_rate": 3.7159187036528426e-06,
+      "loss": 0.512,
+      "r_loss": 0.0,
+      "step": 6170,
+      "steps": 6170
+    },
+    {
+      "epoch": 0.823,
+      "learning_rate": 3.688467453996155e-06,
+      "loss": 0.5244,
+      "r_loss": 0.0,
+      "step": 6180,
+      "steps": 6180
+    },
+    {
+      "epoch": 0.824,
+      "learning_rate": 3.661016204339467e-06,
+      "loss": 0.5224,
+      "r_loss": 0.0,
+      "step": 6190,
+      "steps": 6190
+    },
+    {
+      "epoch": 0.826,
+      "learning_rate": 3.6335649546827792e-06,
+      "loss": 0.5312,
+      "r_loss": 0.0,
+      "step": 6200,
+      "steps": 6200
+    },
+    {
+      "epoch": 0.827,
+      "learning_rate": 3.6061137050260917e-06,
+      "loss": 0.5334,
+      "r_loss": 0.0,
+      "step": 6210,
+      "steps": 6210
+    },
+    {
+      "epoch": 0.828,
+      "learning_rate": 3.5786624553694038e-06,
+      "loss": 0.5401,
+      "r_loss": 0.0,
+      "step": 6220,
+      "steps": 6220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5512112057127163e-06,
+      "loss": 0.5324,
+      "r_loss": 0.0,
+      "step": 6230,
+      "steps": 6230
+    },
+    {
+      "epoch": 0.831,
+      "learning_rate": 3.5237599560560283e-06,
+      "loss": 0.5317,
+      "r_loss": 0.0,
+      "step": 6240,
+      "steps": 6240
+    },
+    {
+      "epoch": 0.832,
+      "learning_rate": 3.4963087063993404e-06,
+      "loss": 0.5368,
+      "r_loss": 0.0,
+      "step": 6250,
+      "steps": 6250
+    },
+    {
+      "epoch": 0.834,
+      "learning_rate": 3.4688574567426533e-06,
+      "loss": 0.5441,
+      "r_loss": 0.0,
+      "step": 6260,
+      "steps": 6260
+    },
+    {
+      "epoch": 0.835,
+      "learning_rate": 3.4414062070859653e-06,
+      "loss": 0.5257,
+      "r_loss": 0.0,
+      "step": 6270,
+      "steps": 6270
+    },
+    {
+      "epoch": 0.836,
+      "learning_rate": 3.413954957429278e-06,
+      "loss": 0.535,
+      "r_loss": 0.0,
+      "step": 6280,
+      "steps": 6280
+    },
+    {
+      "epoch": 0.838,
+      "learning_rate": 3.38650370777259e-06,
+      "loss": 0.5289,
+      "r_loss": 0.0,
+      "step": 6290,
+      "steps": 6290
+    },
+    {
+      "epoch": 0.839,
+      "learning_rate": 3.359052458115902e-06,
+      "loss": 0.534,
+      "r_loss": 0.0,
+      "step": 6300,
+      "steps": 6300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3316012084592144e-06,
+      "loss": 0.5325,
+      "r_loss": 0.0,
+      "step": 6310,
+      "steps": 6310
+    },
+    {
+      "epoch": 0.842,
+      "learning_rate": 3.3041499588025265e-06,
+      "loss": 0.513,
+      "r_loss": 0.0,
+      "step": 6320,
+      "steps": 6320
+    },
+    {
+      "epoch": 0.843,
+      "learning_rate": 3.2766987091458394e-06,
+      "loss": 0.5259,
+      "r_loss": 0.0,
+      "step": 6330,
+      "steps": 6330
+    },
+    {
+      "epoch": 0.844,
+      "learning_rate": 3.249247459489151e-06,
+      "loss": 0.5192,
+      "r_loss": 0.0,
+      "step": 6340,
+      "steps": 6340
+    },
+    {
+      "epoch": 0.846,
+      "learning_rate": 3.221796209832463e-06,
+      "loss": 0.53,
+      "r_loss": 0.0,
+      "step": 6350,
+      "steps": 6350
+    },
+    {
+      "epoch": 0.847,
+      "learning_rate": 3.194344960175776e-06,
+      "loss": 0.5325,
+      "r_loss": 0.0,
+      "step": 6360,
+      "steps": 6360
+    },
+    {
+      "epoch": 0.848,
+      "learning_rate": 3.166893710519088e-06,
+      "loss": 0.5253,
+      "r_loss": 0.0,
+      "step": 6370,
+      "steps": 6370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1394424608624005e-06,
+      "loss": 0.5233,
+      "r_loss": 0.0,
+      "step": 6380,
+      "steps": 6380
+    },
+    {
+      "epoch": 0.851,
+      "learning_rate": 3.1119912112057125e-06,
+      "loss": 0.522,
+      "r_loss": 0.0,
+      "step": 6390,
+      "steps": 6390
+    },
+    {
+      "epoch": 0.852,
+      "learning_rate": 3.084539961549025e-06,
+      "loss": 0.5424,
+      "r_loss": 0.0,
+      "step": 6400,
+      "steps": 6400
+    },
+    {
+      "epoch": 0.854,
+      "learning_rate": 3.057088711892337e-06,
+      "loss": 0.5301,
+      "r_loss": 0.0,
+      "step": 6410,
+      "steps": 6410
+    },
+    {
+      "epoch": 0.855,
+      "learning_rate": 3.029637462235649e-06,
+      "loss": 0.5258,
+      "r_loss": 0.0,
+      "step": 6420,
+      "steps": 6420
+    },
+    {
+      "epoch": 0.856,
+      "learning_rate": 3.002186212578962e-06,
+      "loss": 0.5255,
+      "r_loss": 0.0,
+      "step": 6430,
+      "steps": 6430
+    },
+    {
+      "epoch": 0.858,
+      "learning_rate": 2.974734962922274e-06,
+      "loss": 0.5303,
+      "r_loss": 0.0,
+      "step": 6440,
+      "steps": 6440
+    },
+    {
+      "epoch": 0.859,
+      "learning_rate": 2.9472837132655866e-06,
+      "loss": 0.5293,
+      "r_loss": 0.0,
+      "step": 6450,
+      "steps": 6450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9198324636088986e-06,
+      "loss": 0.5339,
+      "r_loss": 0.0,
+      "step": 6460,
+      "steps": 6460
+    },
+    {
+      "epoch": 0.862,
+      "learning_rate": 2.8923812139522107e-06,
+      "loss": 0.5175,
+      "r_loss": 0.0,
+      "step": 6470,
+      "steps": 6470
+    },
+    {
+      "epoch": 0.863,
+      "learning_rate": 2.864929964295523e-06,
+      "loss": 0.5302,
+      "r_loss": 0.0,
+      "step": 6480,
+      "steps": 6480
+    },
+    {
+      "epoch": 0.864,
+      "learning_rate": 2.8374787146388352e-06,
+      "loss": 0.5434,
+      "r_loss": 0.0,
+      "step": 6490,
+      "steps": 6490
+    },
+    {
+      "epoch": 0.866,
+      "learning_rate": 2.8100274649821477e-06,
+      "loss": 0.5311,
+      "r_loss": 0.0,
+      "step": 6500,
+      "steps": 6500
+    },
+    {
+      "epoch": 0.867,
+      "learning_rate": 2.7825762153254598e-06,
+      "loss": 0.5399,
+      "r_loss": 0.0,
+      "step": 6510,
+      "steps": 6510
+    },
+    {
+      "epoch": 0.868,
+      "learning_rate": 2.755124965668772e-06,
+      "loss": 0.5403,
+      "r_loss": 0.0,
+      "step": 6520,
+      "steps": 6520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7276737160120847e-06,
+      "loss": 0.518,
+      "r_loss": 0.0,
+      "step": 6530,
+      "steps": 6530
+    },
+    {
+      "epoch": 0.871,
+      "learning_rate": 2.7002224663553968e-06,
+      "loss": 0.5257,
+      "r_loss": 0.0,
+      "step": 6540,
+      "steps": 6540
+    },
+    {
+      "epoch": 0.872,
+      "learning_rate": 2.6727712166987093e-06,
+      "loss": 0.5383,
+      "r_loss": 0.0,
+      "step": 6550,
+      "steps": 6550
+    },
+    {
+      "epoch": 0.874,
+      "learning_rate": 2.6453199670420213e-06,
+      "loss": 0.5359,
+      "r_loss": 0.0,
+      "step": 6560,
+      "steps": 6560
+    },
+    {
+      "epoch": 0.875,
+      "learning_rate": 2.6178687173853334e-06,
+      "loss": 0.5275,
+      "r_loss": 0.0,
+      "step": 6570,
+      "steps": 6570
+    },
+    {
+      "epoch": 0.876,
+      "learning_rate": 2.590417467728646e-06,
+      "loss": 0.5237,
+      "r_loss": 0.0,
+      "step": 6580,
+      "steps": 6580
+    },
+    {
+      "epoch": 0.878,
+      "learning_rate": 2.562966218071958e-06,
+      "loss": 0.5238,
+      "r_loss": 0.0,
+      "step": 6590,
+      "steps": 6590
+    },
+    {
+      "epoch": 0.879,
+      "learning_rate": 2.535514968415271e-06,
+      "loss": 0.5244,
+      "r_loss": 0.0,
+      "step": 6600,
+      "steps": 6600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.508063718758583e-06,
+      "loss": 0.5351,
+      "r_loss": 0.0,
+      "step": 6610,
+      "steps": 6610
+    },
+    {
+      "epoch": 0.882,
+      "learning_rate": 2.480612469101895e-06,
+      "loss": 0.5365,
+      "r_loss": 0.0,
+      "step": 6620,
+      "steps": 6620
+    },
+    {
+      "epoch": 0.883,
+      "learning_rate": 2.4531612194452074e-06,
+      "loss": 0.5302,
+      "r_loss": 0.0,
+      "step": 6630,
+      "steps": 6630
+    },
+    {
+      "epoch": 0.884,
+      "learning_rate": 2.42570996978852e-06,
+      "loss": 0.5254,
+      "r_loss": 0.0,
+      "step": 6640,
+      "steps": 6640
+    },
+    {
+      "epoch": 0.886,
+      "learning_rate": 2.398258720131832e-06,
+      "loss": 0.5339,
+      "r_loss": 0.0,
+      "step": 6650,
+      "steps": 6650
+    },
+    {
+      "epoch": 0.887,
+      "learning_rate": 2.370807470475144e-06,
+      "loss": 0.5343,
+      "r_loss": 0.0,
+      "step": 6660,
+      "steps": 6660
+    },
+    {
+      "epoch": 0.888,
+      "learning_rate": 2.3433562208184565e-06,
+      "loss": 0.5367,
+      "r_loss": 0.0,
+      "step": 6670,
+      "steps": 6670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3159049711617685e-06,
+      "loss": 0.5354,
+      "r_loss": 0.0,
+      "step": 6680,
+      "steps": 6680
+    },
+    {
+      "epoch": 0.891,
+      "learning_rate": 2.288453721505081e-06,
+      "loss": 0.5185,
+      "r_loss": 0.0,
+      "step": 6690,
+      "steps": 6690
+    },
+    {
+      "epoch": 0.892,
+      "learning_rate": 2.261002471848393e-06,
+      "loss": 0.5343,
+      "r_loss": 0.0,
+      "step": 6700,
+      "steps": 6700
+    },
+    {
+      "epoch": 0.894,
+      "learning_rate": 2.2335512221917056e-06,
+      "loss": 0.5156,
+      "r_loss": 0.0,
+      "step": 6710,
+      "steps": 6710
+    },
+    {
+      "epoch": 0.895,
+      "learning_rate": 2.2060999725350176e-06,
+      "loss": 0.5254,
+      "r_loss": 0.0,
+      "step": 6720,
+      "steps": 6720
+    },
+    {
+      "epoch": 0.896,
+      "learning_rate": 2.17864872287833e-06,
+      "loss": 0.5123,
+      "r_loss": 0.0,
+      "step": 6730,
+      "steps": 6730
+    },
+    {
+      "epoch": 0.898,
+      "learning_rate": 2.1511974732216426e-06,
+      "loss": 0.5281,
+      "r_loss": 0.0,
+      "step": 6740,
+      "steps": 6740
+    },
+    {
+      "epoch": 0.899,
+      "learning_rate": 2.1237462235649546e-06,
+      "loss": 0.5234,
+      "r_loss": 0.0,
+      "step": 6750,
+      "steps": 6750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0962949739082667e-06,
+      "loss": 0.5221,
+      "r_loss": 0.0,
+      "step": 6760,
+      "steps": 6760
+    },
+    {
+      "epoch": 0.902,
+      "learning_rate": 2.068843724251579e-06,
+      "loss": 0.5177,
+      "r_loss": 0.0,
+      "step": 6770,
+      "steps": 6770
+    },
+    {
+      "epoch": 0.903,
+      "learning_rate": 2.0413924745948916e-06,
+      "loss": 0.5225,
+      "r_loss": 0.0,
+      "step": 6780,
+      "steps": 6780
+    },
+    {
+      "epoch": 0.904,
+      "learning_rate": 2.0139412249382037e-06,
+      "loss": 0.531,
+      "r_loss": 0.0,
+      "step": 6790,
+      "steps": 6790
+    },
+    {
+      "epoch": 0.906,
+      "learning_rate": 1.986489975281516e-06,
+      "loss": 0.5303,
+      "r_loss": 0.0,
+      "step": 6800,
+      "steps": 6800
+    },
+    {
+      "epoch": 0.907,
+      "learning_rate": 1.9590387256248282e-06,
+      "loss": 0.5236,
+      "r_loss": 0.0,
+      "step": 6810,
+      "steps": 6810
+    },
+    {
+      "epoch": 0.908,
+      "learning_rate": 1.9315874759681403e-06,
+      "loss": 0.5192,
+      "r_loss": 0.0,
+      "step": 6820,
+      "steps": 6820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9041362263114528e-06,
+      "loss": 0.5255,
+      "r_loss": 0.0,
+      "step": 6830,
+      "steps": 6830
+    },
+    {
+      "epoch": 0.911,
+      "learning_rate": 1.8766849766547652e-06,
+      "loss": 0.5121,
+      "r_loss": 0.0,
+      "step": 6840,
+      "steps": 6840
+    },
+    {
+      "epoch": 0.912,
+      "learning_rate": 1.8492337269980775e-06,
+      "loss": 0.5227,
+      "r_loss": 0.0,
+      "step": 6850,
+      "steps": 6850
+    },
+    {
+      "epoch": 0.914,
+      "learning_rate": 1.8217824773413896e-06,
+      "loss": 0.5183,
+      "r_loss": 0.0,
+      "step": 6860,
+      "steps": 6860
+    },
+    {
+      "epoch": 0.915,
+      "learning_rate": 1.7943312276847018e-06,
+      "loss": 0.5269,
+      "r_loss": 0.0,
+      "step": 6870,
+      "steps": 6870
+    },
+    {
+      "epoch": 0.916,
+      "learning_rate": 1.7668799780280141e-06,
+      "loss": 0.5287,
+      "r_loss": 0.0,
+      "step": 6880,
+      "steps": 6880
+    },
+    {
+      "epoch": 0.918,
+      "learning_rate": 1.7394287283713266e-06,
+      "loss": 0.519,
+      "r_loss": 0.0,
+      "step": 6890,
+      "steps": 6890
+    },
+    {
+      "epoch": 0.919,
+      "learning_rate": 1.7119774787146389e-06,
+      "loss": 0.5352,
+      "r_loss": 0.0,
+      "step": 6900,
+      "steps": 6900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.684526229057951e-06,
+      "loss": 0.5184,
+      "r_loss": 0.0,
+      "step": 6910,
+      "steps": 6910
+    },
+    {
+      "epoch": 0.922,
+      "learning_rate": 1.6570749794012632e-06,
+      "loss": 0.524,
+      "r_loss": 0.0,
+      "step": 6920,
+      "steps": 6920
+    },
+    {
+      "epoch": 0.923,
+      "learning_rate": 1.6296237297445754e-06,
+      "loss": 0.53,
+      "r_loss": 0.0,
+      "step": 6930,
+      "steps": 6930
+    },
+    {
+      "epoch": 0.924,
+      "learning_rate": 1.602172480087888e-06,
+      "loss": 0.5336,
+      "r_loss": 0.0,
+      "step": 6940,
+      "steps": 6940
+    },
+    {
+      "epoch": 0.926,
+      "learning_rate": 1.5747212304312002e-06,
+      "loss": 0.5254,
+      "r_loss": 0.0,
+      "step": 6950,
+      "steps": 6950
+    },
+    {
+      "epoch": 0.927,
+      "learning_rate": 1.5472699807745125e-06,
+      "loss": 0.5217,
+      "r_loss": 0.0,
+      "step": 6960,
+      "steps": 6960
+    },
+    {
+      "epoch": 0.928,
+      "learning_rate": 1.5198187311178245e-06,
+      "loss": 0.5219,
+      "r_loss": 0.0,
+      "step": 6970,
+      "steps": 6970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.492367481461137e-06,
+      "loss": 0.5325,
+      "r_loss": 0.0,
+      "step": 6980,
+      "steps": 6980
+    },
+    {
+      "epoch": 0.931,
+      "learning_rate": 1.4649162318044493e-06,
+      "loss": 0.5206,
+      "r_loss": 0.0,
+      "step": 6990,
+      "steps": 6990
+    },
+    {
+      "epoch": 0.932,
+      "learning_rate": 1.4374649821477615e-06,
+      "loss": 0.5361,
+      "r_loss": 0.0,
+      "step": 7000,
+      "steps": 7000
+    },
+    {
+      "epoch": 0.934,
+      "learning_rate": 1.4100137324910738e-06,
+      "loss": 0.5135,
+      "r_loss": 0.0,
+      "step": 7010,
+      "steps": 7010
+    },
+    {
+      "epoch": 0.935,
+      "learning_rate": 1.3825624828343859e-06,
+      "loss": 0.5374,
+      "r_loss": 0.0,
+      "step": 7020,
+      "steps": 7020
+    },
+    {
+      "epoch": 0.936,
+      "learning_rate": 1.3551112331776983e-06,
+      "loss": 0.524,
+      "r_loss": 0.0,
+      "step": 7030,
+      "steps": 7030
+    },
+    {
+      "epoch": 0.938,
+      "learning_rate": 1.3276599835210106e-06,
+      "loss": 0.5217,
+      "r_loss": 0.0,
+      "step": 7040,
+      "steps": 7040
+    },
+    {
+      "epoch": 0.939,
+      "learning_rate": 1.3002087338643229e-06,
+      "loss": 0.5291,
+      "r_loss": 0.0,
+      "step": 7050,
+      "steps": 7050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2727574842076354e-06,
+      "loss": 0.5251,
+      "r_loss": 0.0,
+      "step": 7060,
+      "steps": 7060
+    },
+    {
+      "epoch": 0.942,
+      "learning_rate": 1.2453062345509474e-06,
+      "loss": 0.5211,
+      "r_loss": 0.0,
+      "step": 7070,
+      "steps": 7070
+    },
+    {
+      "epoch": 0.943,
+      "learning_rate": 1.21785498489426e-06,
+      "loss": 0.5112,
+      "r_loss": 0.0,
+      "step": 7080,
+      "steps": 7080
+    },
+    {
+      "epoch": 0.944,
+      "learning_rate": 1.190403735237572e-06,
+      "loss": 0.5235,
+      "r_loss": 0.0,
+      "step": 7090,
+      "steps": 7090
+    },
+    {
+      "epoch": 0.946,
+      "learning_rate": 1.1629524855808842e-06,
+      "loss": 0.5224,
+      "r_loss": 0.0,
+      "step": 7100,
+      "steps": 7100
+    },
+    {
+      "epoch": 0.947,
+      "learning_rate": 1.1355012359241965e-06,
+      "loss": 0.5225,
+      "r_loss": 0.0,
+      "step": 7110,
+      "steps": 7110
+    },
+    {
+      "epoch": 0.948,
+      "learning_rate": 1.1080499862675088e-06,
+      "loss": 0.5262,
+      "r_loss": 0.0,
+      "step": 7120,
+      "steps": 7120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0805987366108212e-06,
+      "loss": 0.5164,
+      "r_loss": 0.0,
+      "step": 7130,
+      "steps": 7130
+    },
+    {
+      "epoch": 0.951,
+      "learning_rate": 1.0531474869541333e-06,
+      "loss": 0.5273,
+      "r_loss": 0.0,
+      "step": 7140,
+      "steps": 7140
+    },
+    {
+      "epoch": 0.952,
+      "learning_rate": 1.0256962372974458e-06,
+      "loss": 0.5264,
+      "r_loss": 0.0,
+      "step": 7150,
+      "steps": 7150
+    },
+    {
+      "epoch": 0.954,
+      "learning_rate": 9.98244987640758e-07,
+      "loss": 0.5396,
+      "r_loss": 0.0,
+      "step": 7160,
+      "steps": 7160
+    },
+    {
+      "epoch": 0.955,
+      "learning_rate": 9.7079373798407e-07,
+      "loss": 0.5188,
+      "r_loss": 0.0,
+      "step": 7170,
+      "steps": 7170
+    },
+    {
+      "epoch": 0.956,
+      "learning_rate": 9.433424883273827e-07,
+      "loss": 0.5253,
+      "r_loss": 0.0,
+      "step": 7180,
+      "steps": 7180
+    },
+    {
+      "epoch": 0.958,
+      "learning_rate": 9.158912386706948e-07,
+      "loss": 0.5289,
+      "r_loss": 0.0,
+      "step": 7190,
+      "steps": 7190
+    },
+    {
+      "epoch": 0.959,
+      "learning_rate": 8.884399890140071e-07,
+      "loss": 0.5341,
+      "r_loss": 0.0,
+      "step": 7200,
+      "steps": 7200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.609887393573195e-07,
+      "loss": 0.5255,
+      "r_loss": 0.0,
+      "step": 7210,
+      "steps": 7210
+    },
+    {
+      "epoch": 0.962,
+      "learning_rate": 8.335374897006316e-07,
+      "loss": 0.5181,
+      "r_loss": 0.0,
+      "step": 7220,
+      "steps": 7220
+    },
+    {
+      "epoch": 0.963,
+      "learning_rate": 8.06086240043944e-07,
+      "loss": 0.5258,
+      "r_loss": 0.0,
+      "step": 7230,
+      "steps": 7230
+    },
+    {
+      "epoch": 0.964,
+      "learning_rate": 7.786349903872563e-07,
+      "loss": 0.5237,
+      "r_loss": 0.0,
+      "step": 7240,
+      "steps": 7240
+    },
+    {
+      "epoch": 0.966,
+      "learning_rate": 7.511837407305686e-07,
+      "loss": 0.5165,
+      "r_loss": 0.0,
+      "step": 7250,
+      "steps": 7250
+    },
+    {
+      "epoch": 0.967,
+      "learning_rate": 7.237324910738808e-07,
+      "loss": 0.5336,
+      "r_loss": 0.0,
+      "step": 7260,
+      "steps": 7260
+    },
+    {
+      "epoch": 0.968,
+      "learning_rate": 6.96281241417193e-07,
+      "loss": 0.5271,
+      "r_loss": 0.0,
+      "step": 7270,
+      "steps": 7270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.688299917605054e-07,
+      "loss": 0.5185,
+      "r_loss": 0.0,
+      "step": 7280,
+      "steps": 7280
+    },
+    {
+      "epoch": 0.971,
+      "learning_rate": 6.413787421038177e-07,
+      "loss": 0.5241,
+      "r_loss": 0.0,
+      "step": 7290,
+      "steps": 7290
+    },
+    {
+      "epoch": 0.972,
+      "learning_rate": 6.1392749244713e-07,
+      "loss": 0.5139,
+      "r_loss": 0.0,
+      "step": 7300,
+      "steps": 7300
+    },
+    {
+      "epoch": 0.974,
+      "learning_rate": 5.864762427904422e-07,
+      "loss": 0.5298,
+      "r_loss": 0.0,
+      "step": 7310,
+      "steps": 7310
+    },
+    {
+      "epoch": 0.975,
+      "learning_rate": 5.590249931337544e-07,
+      "loss": 0.5104,
+      "r_loss": 0.0,
+      "step": 7320,
+      "steps": 7320
+    },
+    {
+      "epoch": 0.976,
+      "learning_rate": 5.315737434770667e-07,
+      "loss": 0.5294,
+      "r_loss": 0.0,
+      "step": 7330,
+      "steps": 7330
+    },
+    {
+      "epoch": 0.978,
+      "learning_rate": 5.041224938203791e-07,
+      "loss": 0.5149,
+      "r_loss": 0.0,
+      "step": 7340,
+      "steps": 7340
+    },
+    {
+      "epoch": 0.979,
+      "learning_rate": 4.7667124416369135e-07,
+      "loss": 0.5209,
+      "r_loss": 0.0,
+      "step": 7350,
+      "steps": 7350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.4921999450700356e-07,
+      "loss": 0.5275,
+      "r_loss": 0.0,
+      "step": 7360,
+      "steps": 7360
+    },
+    {
+      "epoch": 0.982,
+      "learning_rate": 4.2176874485031583e-07,
+      "loss": 0.5339,
+      "r_loss": 0.0,
+      "step": 7370,
+      "steps": 7370
+    },
+    {
+      "epoch": 0.983,
+      "learning_rate": 3.9431749519362815e-07,
+      "loss": 0.5228,
+      "r_loss": 0.0,
+      "step": 7380,
+      "steps": 7380
+    },
+    {
+      "epoch": 0.984,
+      "learning_rate": 3.668662455369404e-07,
+      "loss": 0.5294,
+      "r_loss": 0.0,
+      "step": 7390,
+      "steps": 7390
+    },
+    {
+      "epoch": 0.986,
+      "learning_rate": 3.394149958802527e-07,
+      "loss": 0.5226,
+      "r_loss": 0.0,
+      "step": 7400,
+      "steps": 7400
+    },
+    {
+      "epoch": 0.987,
+      "learning_rate": 3.11963746223565e-07,
+      "loss": 0.5099,
+      "r_loss": 0.0,
+      "step": 7410,
+      "steps": 7410
+    },
+    {
+      "epoch": 0.988,
+      "learning_rate": 2.845124965668772e-07,
+      "loss": 0.5312,
+      "r_loss": 0.0,
+      "step": 7420,
+      "steps": 7420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5706124691018954e-07,
+      "loss": 0.5155,
+      "r_loss": 0.0,
+      "step": 7430,
+      "steps": 7430
+    },
+    {
+      "epoch": 0.991,
+      "learning_rate": 2.2960999725350179e-07,
+      "loss": 0.5098,
+      "r_loss": 0.0,
+      "step": 7440,
+      "steps": 7440
+    },
+    {
+      "epoch": 0.992,
+      "learning_rate": 2.0215874759681408e-07,
+      "loss": 0.5206,
+      "r_loss": 0.0,
+      "step": 7450,
+      "steps": 7450
+    },
+    {
+      "epoch": 0.994,
+      "learning_rate": 1.7470749794012635e-07,
+      "loss": 0.5252,
+      "r_loss": 0.0,
+      "step": 7460,
+      "steps": 7460
+    },
+    {
+      "epoch": 0.995,
+      "learning_rate": 1.4725624828343862e-07,
+      "loss": 0.5238,
+      "r_loss": 0.0,
+      "step": 7470,
+      "steps": 7470
+    },
+    {
+      "epoch": 0.996,
+      "learning_rate": 1.1980499862675088e-07,
+      "loss": 0.5272,
+      "r_loss": 0.0,
+      "step": 7480,
+      "steps": 7480
+    },
+    {
+      "epoch": 0.998,
+      "learning_rate": 9.235374897006317e-08,
+      "loss": 0.528,
+      "r_loss": 0.0,
+      "step": 7490,
+      "steps": 7490
+    },
+    {
+      "epoch": 0.999,
+      "learning_rate": 6.490249931337545e-08,
+      "loss": 0.5232,
+      "r_loss": 0.0,
+      "step": 7500,
+      "steps": 7500
+    },
+    {
+      "epoch": 1.0,
+      "step": 7508,
+      "steps": 7508,
+      "total_flos": 1.950614777001463e+19,
+      "train_loss": 0.5782644650881508,
+      "train_runtime": 47075.3922,
+      "train_samples_per_second": 20.414,
+      "train_steps_per_second": 0.159
+    }
+  ],
+  "max_steps": 7508,
+  "num_train_epochs": 1,
+  "total_flos": 1.950614777001463e+19,
+  "trial_name": null,
+  "trial_params": null
+}